Files
seaweedfs/weed/command/filer_copy.go
T
Eliah RusinandChris Lu cfa8afec92 filer: guard FoundationDB 100KB value limit and pack earlier (#11161)
* filer: guard the FoundationDB value size limit, not the transaction limit

An entry's whole chunk list is one FoundationDB value, and FDB caps a value at
100,000 bytes while a transaction may reach 10MB. UpdateEntry checked the
transaction limit, so every entry between the two limits passed the guard and
was rejected by FDB itself with error 2103 (Value length exceeds limit). The
failure surfaced inside the store rather than at the guard, so the S3 layer
dropped the connection and clients saw a network fault instead of an error.

Check the value limit in UpdateEntry and KvPut instead, after gzip and before
the transaction, with an error that names the limit it hit. The removed
transaction-size constant guarded nothing else: DeleteFolderChildren batches by
entry count.

Refs #11158

* filer: fold at 500 chunks in the foundationdb build

Manifest packing is what keeps a large file's entry small, but it only ran once
a flat chunk list reached 10000 chunks. A FoundationDB value stops at 100,000
bytes and an entry's whole chunk list is one value, which at ~100 bytes per
chunk record is about 1000 chunks -- so on FDB the write always failed before
packing could help: a 3.3 GiB PutObject at the default -maxMB=4 was already
past the limit.

FoundationDB support is its own build (`go build -tags foundationdb`, shipped
as its own image), so the batch is a build-time choice and needs no negotiation
at run time. The tagged build folds at 500, every other build keeps 10000 and
is untouched.

500 is not arbitrary: a single fold level leaves (chunks/batch) manifest
pointers plus up to (batch-1) unfolded chunks in the entry, so the reachable
chunk count is highest when the two terms are near equal. For a 100,000-byte
budget that optimum is 500, which holds an entry inside the limit up to
~250,000 chunks -- ~1 TB at -maxMB=4, against ~4 GB before. Larger files need
nested packing, which no batch size substitutes for.

One binary serves every role in that image, so the filer and each client that
folds -- S3, mount, WebDAV, weed shell, filer.copy -- agree on the batch by
construction. A binary built with the tag but pointed at another store folds
earlier than that store requires, costing one manifest blob per 500 chunks and
one read to resolve it.

Fixes #11158

* filer: fold with rollback inside MaybeManifestize, not beside it

A fold that fails midway has already uploaded manifest blobs for its earlier
batches, and returns only the data chunks -- dropping the manifests it had
separated out of the caller's list. Both were wrong in ways that mattered:

  - AppendToEntry assigned that shortened list straight to entry.Chunks and
    created the entry, so an append to an already-folded file whose fold
    failed lost every previously folded chunk. weed mount had the same shape.
  - cleanupChunks logged the error as "not good, but should be ok" and then
    returned it through a named result, failing the whole CreateEntry or
    UpdateEntry, while the blobs it had written stayed behind referenced by
    nothing.

The S3 path was alone in handling this, through a private helper beside
MaybeManifestize. A second entry point next to the one everything else calls
just means the wrong one gets used, so the behaviour moves inside
MaybeManifestize: on failure it returns inputChunks as it received them, and
hands the blobs it saved to a deleteChunks callback. The filer, S3 and
filer.copy pass their existing deleters -- filer.copy already cleans up this
way after a failed upload -- and mount, WebDAV and weed shell pass nil, which
reports the blobs rather than collecting them, as before. Each caller keeps its
own error policy: the filer HTTP PUT path and filer.copy still fail the request,
the rest still continue with the flat list, which is a correct entry.

---------

Co-authored-by: Chris Lu <chris.lu@gmail.com>
2026-09-04 23:12:27 -07:00

623 lines
18 KiB
Go

package command
import (
"context"
"fmt"
"io"
"net/http"
"os"
"path/filepath"
"strconv"
"strings"
"sync"
"time"
"google.golang.org/grpc"
"github.com/seaweedfs/seaweedfs/weed/filer"
"github.com/seaweedfs/seaweedfs/weed/operation"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/security"
"github.com/seaweedfs/seaweedfs/weed/storage/needle"
"github.com/seaweedfs/seaweedfs/weed/util"
"github.com/seaweedfs/seaweedfs/weed/util/grace"
)
var (
copyOptions CopyOptions
waitGroup sync.WaitGroup
)
type CopyOptions struct {
include *string
replication *string
collection *string
ttl *string
diskType *string
maxMB *int
concurrentFiles *int
concurrentChunks *int
grpcDialOption grpc.DialOption
masters []string
cipher bool
ttlSec int32
checkSize *bool
verbose *bool
volumeServerAccess *string
}
func init() {
cmdFilerCopy.Run = runCopy // break init cycle
cmdFilerCopy.IsDebug = cmdFilerCopy.Flag.Bool("debug", false, "verbose debug information")
copyOptions.include = cmdFilerCopy.Flag.String("include", "", "patterns of files to copy, e.g., *.pdf, *.html, ab?d.txt, works together with -dir")
copyOptions.replication = cmdFilerCopy.Flag.String("replication", "", "replication type")
copyOptions.collection = cmdFilerCopy.Flag.String("collection", "", "optional collection name")
copyOptions.ttl = cmdFilerCopy.Flag.String("ttl", "", "time to live, e.g.: 1m, 1h, 1d, 1M, 1y")
copyOptions.diskType = cmdFilerCopy.Flag.String("disk", "", "[hdd|ssd|<tag>] hard drive or solid state drive or any tag")
copyOptions.maxMB = cmdFilerCopy.Flag.Int("maxMB", 4, "split files larger than the limit")
copyOptions.concurrentFiles = cmdFilerCopy.Flag.Int("c", 8, "concurrent file copy goroutines")
copyOptions.concurrentChunks = cmdFilerCopy.Flag.Int("concurrentChunks", 8, "concurrent chunk copy goroutines for each file")
copyOptions.checkSize = cmdFilerCopy.Flag.Bool("check.size", false, "copy when the target file size is different from the source file")
copyOptions.verbose = cmdFilerCopy.Flag.Bool("verbose", false, "print out details during copying")
copyOptions.volumeServerAccess = cmdFilerCopy.Flag.String("volumeServerAccess", "direct", "access volume servers by [direct|publicUrl]")
}
var cmdFilerCopy = &Command{
UsageLine: "filer.copy file_or_dir1 [file_or_dir2 file_or_dir3] http://localhost:8888/path/to/a/folder/",
Short: "copy one or a list of files to a filer folder",
Long: `copy one or a list of files, or batch copy one whole folder recursively, to a filer folder
It can copy one or a list of files or folders.
If copying a whole folder recursively:
All files under the folder and sub folders will be copied.
Optional parameter "-include" allows you to specify the file name patterns.
If "maxMB" is set to a positive number, files larger than it would be split into chunks.
`,
}
func runCopy(cmd *Command, args []string) bool {
util.LoadSecurityConfiguration()
if len(args) <= 1 {
return false
}
filerDestination := args[len(args)-1]
fileOrDirs := args[0 : len(args)-1]
filerAddress, urlPath, err := pb.ParseUrl(filerDestination)
if err != nil {
fmt.Printf("The last argument should be a URL on filer: %v\n", err)
return false
}
if !strings.HasSuffix(urlPath, "/") {
fmt.Printf("The last argument should be a folder and end with \"/\"\n")
return false
}
copyOptions.grpcDialOption = security.LoadClientTLS(util.GetViper(), "grpc.client")
masters, collection, replication, dirBuckets, maxMB, cipher, err := readFilerConfiguration(copyOptions.grpcDialOption, filerAddress)
if err != nil {
fmt.Printf("read from filer %s: %v\n", filerAddress, err)
return false
}
if strings.HasPrefix(urlPath, dirBuckets+"/") {
restPath := urlPath[len(dirBuckets)+1:]
if strings.Index(restPath, "/") > 0 {
expectedBucket := restPath[:strings.Index(restPath, "/")]
if *copyOptions.collection == "" {
*copyOptions.collection = expectedBucket
} else if *copyOptions.collection != expectedBucket {
fmt.Printf("destination %s uses collection \"%s\": unexpected collection \"%v\"\n", urlPath, expectedBucket, *copyOptions.collection)
return true
}
}
}
if *copyOptions.collection == "" {
*copyOptions.collection = collection
}
if *copyOptions.replication == "" {
*copyOptions.replication = replication
}
if *copyOptions.maxMB == 0 {
*copyOptions.maxMB = int(maxMB)
}
copyOptions.masters = masters
copyOptions.cipher = cipher
ttl, err := needle.ReadTTL(*copyOptions.ttl)
if err != nil {
fmt.Printf("parsing ttl %s: %v\n", *copyOptions.ttl, err)
return false
}
copyOptions.ttlSec = int32(ttl.Minutes()) * 60
if *cmdFilerCopy.IsDebug {
grace.SetupProfiling("filer.copyOptions.cpu.pprof", "filer.copyOptions.mem.pprof")
}
concurrentFiles := *copyOptions.concurrentFiles
if concurrentFiles <= 0 {
fmt.Fprintf(os.Stderr, "Invalid concurrency %d; using at least 1 worker\n", concurrentFiles)
concurrentFiles = 1
}
*copyOptions.concurrentFiles = concurrentFiles
fileCopyTaskChan := make(chan FileCopyTask, concurrentFiles)
go func() {
defer close(fileCopyTaskChan)
for _, fileOrDir := range fileOrDirs {
if err := genFileCopyTask(fileOrDir, urlPath, fileCopyTaskChan); err != nil {
fmt.Fprintf(os.Stderr, "genFileCopyTask : %v\n", err)
break
}
}
}()
for i := 0; i < concurrentFiles; i++ {
waitGroup.Add(1)
go func() {
defer waitGroup.Done()
worker := FileCopyWorker{
options: &copyOptions,
filerAddress: filerAddress,
signature: util.RandomInt32(),
}
if err := worker.copyFiles(fileCopyTaskChan); err != nil {
fmt.Fprintf(os.Stderr, "copy file error: %v\n", err)
return
}
}()
}
waitGroup.Wait()
return true
}
func readFilerConfiguration(grpcDialOption grpc.DialOption, filerGrpcAddress pb.ServerAddress) (masters []string, collection, replication string, dirBuckets string, maxMB uint32, cipher bool, err error) {
err = pb.WithGrpcFilerClient(false, 0, filerGrpcAddress, grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.GetFilerConfiguration(context.Background(), &filer_pb.GetFilerConfigurationRequest{})
if err != nil {
return fmt.Errorf("get filer %s configuration: %v", filerGrpcAddress, err)
}
masters, collection, replication, maxMB = resp.Masters, resp.Collection, resp.Replication, resp.MaxMb
dirBuckets = resp.DirBuckets
cipher = resp.Cipher
return nil
})
return
}
func genFileCopyTask(fileOrDir string, destPath string, fileCopyTaskChan chan FileCopyTask) error {
fi, err := os.Stat(fileOrDir)
if err != nil {
fmt.Fprintf(os.Stderr, "Error: read file %s: %v\n", fileOrDir, err)
return nil
}
mode := fi.Mode()
uid, gid := util.GetFileUidGid(fi)
fileSize := fi.Size()
if mode.IsDir() {
fileSize = 0
}
fileCopyTaskChan <- FileCopyTask{
sourceLocation: fileOrDir,
destinationUrlPath: destPath,
fileSize: fileSize,
fileMode: fi.Mode(),
uid: uid,
gid: gid,
}
if mode.IsDir() {
files, _ := os.ReadDir(fileOrDir)
for _, subFileOrDir := range files {
cleanedDestDirectory := destPath + fi.Name()
if err = genFileCopyTask(fileOrDir+"/"+subFileOrDir.Name(), cleanedDestDirectory+"/", fileCopyTaskChan); err != nil {
return err
}
}
}
return nil
}
type FileCopyWorker struct {
options *CopyOptions
filerAddress pb.ServerAddress
signature int32
}
func (worker *FileCopyWorker) copyFiles(fileCopyTaskChan chan FileCopyTask) error {
for task := range fileCopyTaskChan {
if err := worker.doEachCopy(task); err != nil {
return err
}
}
return nil
}
type FileCopyTask struct {
sourceLocation string
destinationUrlPath string
fileSize int64
fileMode os.FileMode
uid uint32
gid uint32
}
func (worker *FileCopyWorker) doEachCopy(task FileCopyTask) error {
f, err := os.Open(task.sourceLocation)
if err != nil {
fmt.Printf("Failed to open file %s: %v\n", task.sourceLocation, err)
if _, ok := err.(*os.PathError); ok {
fmt.Printf("skipping %s\n", task.sourceLocation)
return nil
}
return err
}
defer f.Close()
// this is a regular file
if *worker.options.include != "" {
if ok, _ := filepath.Match(*worker.options.include, filepath.Base(task.sourceLocation)); !ok {
return nil
}
}
if shouldCopy, err := worker.checkExistingFileFirst(task, f); err != nil {
return fmt.Errorf("check existing file: %w", err)
} else if !shouldCopy {
if *worker.options.verbose {
fmt.Printf("skipping copied file: %v\n", f.Name())
}
return nil
}
// find the chunk count
chunkSize := int64(*worker.options.maxMB * 1024 * 1024)
chunkCount := 1
if chunkSize > 0 && task.fileSize > chunkSize {
chunkCount = int(task.fileSize/chunkSize) + 1
}
if chunkCount == 1 {
return worker.uploadFileAsOne(task, f)
}
return worker.uploadFileInChunks(task, f, chunkCount, chunkSize)
}
func (worker *FileCopyWorker) checkExistingFileFirst(task FileCopyTask, f *os.File) (shouldCopy bool, err error) {
shouldCopy = true
if !*worker.options.checkSize {
return
}
fileStat, err := f.Stat()
if err != nil {
shouldCopy = false
return
}
err = pb.WithGrpcFilerClient(false, worker.signature, worker.filerAddress, worker.options.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
request := &filer_pb.LookupDirectoryEntryRequest{
Directory: task.destinationUrlPath,
Name: filepath.Base(f.Name()),
}
resp, lookupErr := client.LookupDirectoryEntry(context.Background(), request)
if lookupErr != nil {
// mostly not found error
return nil
}
if fileStat.Size() == int64(filer.FileSize(resp.Entry)) {
shouldCopy = false
}
return nil
})
return
}
func (worker *FileCopyWorker) uploadFileAsOne(task FileCopyTask, f *os.File) error {
// upload the file content
fileName := filepath.Base(f.Name())
var mimeType string
var chunks []*filer_pb.FileChunk
if task.fileMode&os.ModeDir == 0 && task.fileSize > 0 {
mimeType = detectMimeType(f)
data, err := io.ReadAll(f)
if err != nil {
return err
}
uploader, uploaderErr := operation.NewUploader()
if uploaderErr != nil {
return uploaderErr
}
finalFileId, uploadResult, flushErr, _ := uploader.UploadWithRetry(
worker,
&filer_pb.AssignVolumeRequest{
Count: 1,
Replication: *worker.options.replication,
Collection: *worker.options.collection,
TtlSec: worker.options.ttlSec,
DiskType: *worker.options.diskType,
Path: task.destinationUrlPath,
},
&operation.UploadOption{
Filename: fileName,
Cipher: worker.options.cipher,
IsInputCompressed: false,
MimeType: mimeType,
PairMap: nil,
},
util.NewBytesReader(data),
)
if flushErr != nil {
return flushErr
}
chunks = append(chunks, uploadResult.ToPbFileChunk(finalFileId, 0, time.Now().UnixNano()))
}
if err := pb.WithGrpcFilerClient(false, worker.signature, worker.filerAddress, worker.options.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
request := &filer_pb.CreateEntryRequest{
Directory: task.destinationUrlPath,
Entry: &filer_pb.Entry{
Name: fileName,
Attributes: &filer_pb.FuseAttributes{
Crtime: time.Now().Unix(),
Mtime: time.Now().Unix(),
Gid: task.gid,
Uid: task.uid,
FileSize: uint64(task.fileSize),
FileMode: uint32(task.fileMode),
Mime: mimeType,
TtlSec: worker.options.ttlSec,
},
Chunks: chunks,
},
}
if err := filer_pb.CreateEntry(context.Background(), client, request); err != nil {
return fmt.Errorf("update fh: %w", err)
}
return nil
}); err != nil {
return fmt.Errorf("upload data %v to http://%s%s%s: %v\n", fileName, worker.filerAddress.ToHttpAddress(), task.destinationUrlPath, fileName, err)
}
return nil
}
func (worker *FileCopyWorker) uploadFileInChunks(task FileCopyTask, f *os.File, chunkCount int, chunkSize int64) error {
fileName := filepath.Base(f.Name())
mimeType := detectMimeType(f)
chunksChan := make(chan *filer_pb.FileChunk, chunkCount)
concurrentChunks := make(chan struct{}, *worker.options.concurrentChunks)
var wg sync.WaitGroup
var uploadError error
fmt.Printf("uploading %s in %d chunks ...\n", fileName, chunkCount)
for i := int64(0); i < int64(chunkCount) && uploadError == nil; i++ {
wg.Add(1)
concurrentChunks <- struct{}{}
go func(i int64) {
defer func() {
wg.Done()
<-concurrentChunks
}()
uploader, err := operation.NewUploader()
if err != nil {
uploadError = fmt.Errorf("upload data %v: %w\n", fileName, err)
return
}
fileId, uploadResult, err, _ := uploader.UploadWithRetry(
worker,
&filer_pb.AssignVolumeRequest{
Count: 1,
Replication: *worker.options.replication,
Collection: *worker.options.collection,
TtlSec: worker.options.ttlSec,
DiskType: *worker.options.diskType,
Path: task.destinationUrlPath + fileName,
},
&operation.UploadOption{
Filename: fileName + "-" + strconv.FormatInt(i+1, 10),
Cipher: worker.options.cipher,
IsInputCompressed: false,
MimeType: "",
PairMap: nil,
},
io.NewSectionReader(f, i*chunkSize, chunkSize),
)
if err != nil {
uploadError = fmt.Errorf("upload data %v: %w\n", fileName, err)
return
}
if uploadResult.Error != "" {
uploadError = fmt.Errorf("upload %v result: %v\n", fileName, uploadResult.Error)
return
}
chunksChan <- uploadResult.ToPbFileChunk(fileId, i*chunkSize, time.Now().UnixNano())
fmt.Printf("uploaded %s-%d [%d,%d)\n", fileName, i+1, i*chunkSize, i*chunkSize+int64(uploadResult.Size))
}(i)
}
wg.Wait()
close(chunksChan)
var chunks []*filer_pb.FileChunk
for chunk := range chunksChan {
chunks = append(chunks, chunk)
}
if uploadError != nil {
var fileIds []string
for _, chunk := range chunks {
fileIds = append(fileIds, chunk.FileId)
}
if len(worker.options.masters) == 0 {
return fmt.Errorf("upload data %v: %w (cleanup skipped: no masters configured)", fileName, uploadError)
}
operation.DeleteFileIds(func(_ context.Context) pb.ServerAddress {
return pb.ServerAddress(worker.options.masters[0])
}, false, worker.options.grpcDialOption, fileIds)
return uploadError
}
// A fold that fails midway aborts the copy, so the blobs its earlier
// batches wrote go the same way as the chunks of a failed upload.
deleteManifestChunks := func(saved []*filer_pb.FileChunk) {
if len(worker.options.masters) == 0 {
return
}
var fileIds []string
for _, chunk := range saved {
fileIds = append(fileIds, chunk.GetFileIdString())
}
operation.DeleteFileIds(func(_ context.Context) pb.ServerAddress {
return pb.ServerAddress(worker.options.masters[0])
}, false, worker.options.grpcDialOption, fileIds)
}
manifestedChunks, manifestErr := filer.MaybeManifestize(worker.saveDataAsChunk, deleteManifestChunks, chunks)
if manifestErr != nil {
return fmt.Errorf("create manifest: %w", manifestErr)
}
if err := pb.WithGrpcFilerClient(false, worker.signature, worker.filerAddress, worker.options.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
request := &filer_pb.CreateEntryRequest{
Directory: task.destinationUrlPath,
Entry: &filer_pb.Entry{
Name: fileName,
Attributes: &filer_pb.FuseAttributes{
Crtime: time.Now().Unix(),
Mtime: time.Now().Unix(),
Gid: task.gid,
Uid: task.uid,
FileSize: uint64(task.fileSize),
FileMode: uint32(task.fileMode),
Mime: mimeType,
TtlSec: worker.options.ttlSec,
},
Chunks: manifestedChunks,
},
}
if err := filer_pb.CreateEntry(context.Background(), client, request); err != nil {
return fmt.Errorf("update fh: %w", err)
}
return nil
}); err != nil {
return fmt.Errorf("upload data %v to http://%s%s%s: %v\n", fileName, worker.filerAddress.ToHttpAddress(), task.destinationUrlPath, fileName, err)
}
fmt.Printf("copied %s => http://%s%s%s\n", f.Name(), worker.filerAddress.ToHttpAddress(), task.destinationUrlPath, fileName)
return nil
}
func detectMimeType(f *os.File) string {
head := make([]byte, 512)
f.Seek(0, io.SeekStart)
n, err := f.Read(head)
if err == io.EOF {
return ""
}
if err != nil {
fmt.Printf("read head of %v: %v\n", f.Name(), err)
return ""
}
f.Seek(0, io.SeekStart)
mimeType := http.DetectContentType(head[:n])
if mimeType == "application/octet-stream" {
return ""
}
return mimeType
}
func (worker *FileCopyWorker) saveDataAsChunk(reader io.Reader, name string, offset int64, tsNs int64, _ uint64) (chunk *filer_pb.FileChunk, err error) {
uploader, uploaderErr := operation.NewUploader()
if uploaderErr != nil {
return nil, fmt.Errorf("upload data: %w", uploaderErr)
}
finalFileId, uploadResult, flushErr, _ := uploader.UploadWithRetry(
worker,
&filer_pb.AssignVolumeRequest{
Count: 1,
Replication: *worker.options.replication,
Collection: *worker.options.collection,
TtlSec: worker.options.ttlSec,
DiskType: *worker.options.diskType,
Path: name,
},
&operation.UploadOption{
Filename: name,
Cipher: worker.options.cipher,
IsInputCompressed: false,
MimeType: "",
PairMap: nil,
},
reader,
)
if flushErr != nil {
return nil, fmt.Errorf("upload data: %w", flushErr)
}
if uploadResult.Error != "" {
return nil, fmt.Errorf("upload result: %v", uploadResult.Error)
}
return uploadResult.ToPbFileChunk(finalFileId, offset, tsNs), nil
}
var _ = filer_pb.FilerClient(&FileCopyWorker{})
func (worker *FileCopyWorker) WithFilerClient(streamingMode bool, fn func(filer_pb.SeaweedFilerClient) error) (err error) {
filerGrpcAddress := worker.filerAddress.ToGrpcAddress()
err = pb.WithGrpcClient(context.Background(), streamingMode, worker.signature, func(grpcConnection *grpc.ClientConn) error {
client := filer_pb.NewSeaweedFilerClient(grpcConnection)
return fn(client)
}, filerGrpcAddress, false, worker.options.grpcDialOption)
return
}
func (worker *FileCopyWorker) AdjustedUrl(location *filer_pb.Location) string {
if *worker.options.volumeServerAccess == "publicUrl" {
return location.PublicUrl
}
return location.Url
}
func (worker *FileCopyWorker) GetDataCenter() string {
return ""
}