Files
seaweedfs/weed/server/filer_grpc_server_rename.go
T
Chris Luandbruce-zzz 0f05957bc4 filer: self-heal chunk manifest reads when volume locations go stale (#11107)
* filer: self-heal fetchWholeChunk on stale volume locations

Upstream #10156/#10800 wired cache invalidation into the buffer-based
read paths, but manifest resolution still goes through fetchWholeChunk,
which returns the raw error on failure. When cached volume locations
are stale (volume tiered to remote storage, server rolled), resolving
a large multipart file fails permanently even though other locations
are healthy.

Thread the ChunkGroup's cacheInvalidator through ResolveChunkManifest /
ResolveOneChunkManifest / fetchWholeChunk, and on failure invalidate,
re-lookup and retry once via the existing retryFetchWithFreshLocations
helper. The streaming bytesBuffer is reset before the retry so partial
bytes from the failed attempt cannot corrupt the manifest
proto.Unmarshal. Non-mount callers pass nil and keep their semantics.

* filer: move the manifest self-heal tests in with the other manifest tests

Also make the stale server stream a prefix and then abort mid-body, which is
what actually leaves partial bytes in the buffer: an HTTP error status returns
before ReadUrlAsStream ever calls the writer, so a 500 never exercised the
Reset the tests claimed to cover.

Claude-Session: https://claude.ai/code/session_01FK3oGC5ZVeJYvNBWgb9JUD

* filer: keep the cached volume locations when a manifest read is cancelled

A cancelled or timed-out read says nothing about where the volume lives, so
dropping the location and going back to the master only costs the next reader
a round trip. PrepareStreamContentWithThrottler already guards its self-heal
this way. The guard also goes inside retryFetchWithFreshLocations, since the
caller can be cancelled between its own check and the invalidation, and that
covers the reader cache and prefetch paths too.

fetchWholeChunk returns the context error rather than the stream failure it
provoked, and ResolveOneChunkManifest wraps with %w so errors.Is still sees it.
That matters even where no invalidator is passed: volume.fsck resolves
manifests with nil and tells its own abort from a corrupt manifest that way,
so the cancellation check sits ahead of the nil-invalidator return.

Claude-Session: https://claude.ai/code/session_01FK3oGC5ZVeJYvNBWgb9JUD

* filer: self-heal manifest reads on the filer and s3 paths too

Every caller that already holds the location cache backing its lookup function
can hand it over: the filer's read, copy and deletion paths and the log cache
have the MasterClient right there, and s3api has the FilerClient. MinusChunks
takes one for the same reason, since the deletion path resolves manifests
through it. Only the shell tools and the replication sinks, whose lookup
functions cache privately with nothing to invalidate, keep passing nil.

Claude-Session: https://claude.ai/code/session_01FK3oGC5ZVeJYvNBWgb9JUD

---------

Co-authored-by: bruce-zzz <bruce.zou@hhy-data.com>
2026-09-02 17:43:46 -07:00

352 lines
12 KiB
Go

package weed_server
import (
"context"
"fmt"
"path/filepath"
"time"
"github.com/seaweedfs/seaweedfs/weed/filer"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/util"
)
// acquireRenamePathLocks holds both paths across commit and notification so a
// fenced lookup cannot read renamed state under a fence preceding its events.
// Path order avoids deadlock with a reverse rename; descendants are not
// locked. The returned func releases both.
func (fs *FilerServer) acquireRenamePathLocks(intention string, oldPath, newPath util.FullPath) func() {
firstPath, secondPath := oldPath, newPath
if secondPath < firstPath {
firstPath, secondPath = secondPath, firstPath
}
firstLock := fs.entryLockTable.AcquireLock(intention, firstPath, util.ExclusiveLock)
if secondPath == firstPath {
return func() { fs.entryLockTable.ReleaseLock(firstPath, firstLock) }
}
secondLock := fs.entryLockTable.AcquireLock(intention, secondPath, util.ExclusiveLock)
return func() {
fs.entryLockTable.ReleaseLock(secondPath, secondLock)
fs.entryLockTable.ReleaseLock(firstPath, firstLock)
}
}
func (fs *FilerServer) AtomicRenameEntry(ctx context.Context, req *filer_pb.AtomicRenameEntryRequest) (*filer_pb.AtomicRenameEntryResponse, error) {
glog.V(1).Infof("AtomicRenameEntry %v", req)
oldParent := util.FullPath(filepath.ToSlash(req.OldDirectory))
newParent := util.FullPath(filepath.ToSlash(req.NewDirectory))
if err := fs.filer.CanRename(ctx, oldParent, newParent, req.OldName); err != nil {
return nil, err
}
defer fs.acquireRenamePathLocks("AtomicRenameEntry", oldParent.Child(req.OldName), newParent.Child(req.NewName))()
ctx, err := fs.filer.BeginTransaction(ctx)
if err != nil {
return nil, err
}
oldEntry, err := fs.filer.FindEntry(ctx, oldParent.Child(req.OldName))
if err != nil {
fs.filer.RollbackTransaction(ctx)
return nil, fmt.Errorf("%s/%s not found: %v", req.OldDirectory, req.OldName, err)
}
var metadataEvents []metadataEvent
var pendingChunkDeletes []*filer_pb.FileChunk
moveErr := fs.moveEntry(ctx, nil, oldParent, oldEntry, newParent, req.NewName, req.Signatures, false, &metadataEvents, &pendingChunkDeletes)
if moveErr != nil {
fs.filer.RollbackTransaction(ctx)
return nil, fmt.Errorf("%s/%s move error: %v", req.OldDirectory, req.OldName, moveErr)
} else {
if commitError := fs.filer.CommitTransaction(ctx); commitError != nil {
fs.filer.RollbackTransaction(ctx)
return nil, fmt.Errorf("%s/%s move commit error: %v", req.OldDirectory, req.OldName, commitError)
}
}
// Chunks from an overwritten rename target are only deletable after the
// rename transaction has committed: anything that fails mid-rename (move,
// child moves, oldPath delete, CommitTransaction) would otherwise leave
// live metadata pointing at freshly-deleted chunks.
if len(pendingChunkDeletes) > 0 {
fs.filer.DeleteChunksNotRecursive(pendingChunkDeletes)
}
for _, event := range metadataEvents {
event.notify(fs.filer, ctx, req.Signatures)
}
return &filer_pb.AtomicRenameEntryResponse{}, nil
}
func (fs *FilerServer) StreamRenameEntry(req *filer_pb.StreamRenameEntryRequest, stream filer_pb.SeaweedFiler_StreamRenameEntryServer) (err error) {
glog.V(1).Infof("StreamRenameEntry %v", req)
oldParent := util.FullPath(filepath.ToSlash(req.OldDirectory))
newParent := util.FullPath(filepath.ToSlash(req.NewDirectory))
if err := fs.filer.CanRename(stream.Context(), oldParent, newParent, req.OldName); err != nil {
return err
}
defer fs.acquireRenamePathLocks("StreamRenameEntry", oldParent.Child(req.OldName), newParent.Child(req.NewName))()
ctx := context.Background()
ctx, err = fs.filer.BeginTransaction(ctx)
if err != nil {
return err
}
oldEntry, err := fs.filer.FindEntry(ctx, oldParent.Child(req.OldName))
if err != nil {
fs.filer.RollbackTransaction(ctx)
return fmt.Errorf("%s/%s not found: %v", req.OldDirectory, req.OldName, err)
}
if oldEntry.IsDirectory() {
// follow https://pubs.opengroup.org/onlinepubs/000095399/functions/rename.html
targetDir := newParent.Child(req.NewName)
newEntry, err := fs.filer.FindEntry(ctx, targetDir)
if err == nil {
if !newEntry.IsDirectory() {
fs.filer.RollbackTransaction(ctx)
return fmt.Errorf("%s is not directory", targetDir)
}
if entries, _, _ := fs.filer.ListDirectoryEntries(context.Background(), targetDir, "", false, 1, "", "", ""); len(entries) > 0 {
return fmt.Errorf("%s is not empty", targetDir)
}
}
}
var metadataEvents []metadataEvent
var pendingChunkDeletes []*filer_pb.FileChunk
moveErr := fs.moveEntry(ctx, stream, oldParent, oldEntry, newParent, req.NewName, req.Signatures, false, &metadataEvents, &pendingChunkDeletes)
if moveErr != nil {
fs.filer.RollbackTransaction(ctx)
return fmt.Errorf("%s/%s move error: %v", req.OldDirectory, req.OldName, moveErr)
} else {
if commitError := fs.filer.CommitTransaction(ctx); commitError != nil {
fs.filer.RollbackTransaction(ctx)
return fmt.Errorf("%s/%s move commit error: %v", req.OldDirectory, req.OldName, commitError)
}
}
if len(pendingChunkDeletes) > 0 {
fs.filer.DeleteChunksNotRecursive(pendingChunkDeletes)
}
for _, event := range metadataEvents {
event.notify(fs.filer, ctx, req.Signatures)
}
return nil
}
type metadataEvent struct {
oldEntry *filer.Entry
newEntry *filer.Entry
deleteChunks bool
}
func (event metadataEvent) notify(f *filer.Filer, ctx context.Context, signatures []int32) {
f.NotifyUpdateEvent(ctx, event.oldEntry, event.newEntry, event.deleteChunks, false, signatures)
}
func (fs *FilerServer) moveEntry(ctx context.Context, stream filer_pb.SeaweedFiler_StreamRenameEntryServer, oldParent util.FullPath, entry *filer.Entry, newParent util.FullPath, newName string, signatures []int32, skipTargetLookup bool, metadataEvents *[]metadataEvent, pendingChunkDeletes *[]*filer_pb.FileChunk) error {
if err := fs.moveSelfEntry(ctx, stream, oldParent, entry, newParent, newName, func() error {
if entry.IsDirectory() {
if err := fs.moveFolderSubEntries(ctx, stream, oldParent, entry, newParent, newName, signatures, metadataEvents, pendingChunkDeletes); err != nil {
return err
}
}
return nil
}, signatures, skipTargetLookup, metadataEvents, pendingChunkDeletes); err != nil {
return fmt.Errorf("fail to move %s => %s: %v", oldParent.Child(entry.Name()), newParent.Child(newName), err)
}
return nil
}
func (fs *FilerServer) moveFolderSubEntries(ctx context.Context, stream filer_pb.SeaweedFiler_StreamRenameEntryServer, oldParent util.FullPath, entry *filer.Entry, newParent util.FullPath, newName string, signatures []int32, metadataEvents *[]metadataEvent, pendingChunkDeletes *[]*filer_pb.FileChunk) error {
currentDirPath := oldParent.Child(entry.Name())
newDirPath := newParent.Child(newName)
glog.V(1).Infof("moving folder %s => %s", currentDirPath, newDirPath)
lastFileName := ""
includeLastFile := false
for {
entries, hasMore, err := fs.filer.ListDirectoryEntries(ctx, currentDirPath, lastFileName, includeLastFile, 1024, "", "", "")
if err != nil {
return err
}
// println("found", len(entries), "entries under", currentDirPath)
for _, item := range entries {
lastFileName = item.Name()
// println("processing", lastFileName)
newChildPath := newDirPath.Child(item.Name())
skipTarget := fs.filer.Store.SameActualStore(newDirPath, newChildPath)
err := fs.moveEntry(ctx, stream, currentDirPath, item, newDirPath, item.Name(), signatures, skipTarget, metadataEvents, pendingChunkDeletes)
if err != nil {
return err
}
}
if !hasMore {
break
}
}
return nil
}
func (fs *FilerServer) moveSelfEntry(ctx context.Context, stream filer_pb.SeaweedFiler_StreamRenameEntryServer, oldParent util.FullPath, entry *filer.Entry, newParent util.FullPath, newName string, moveFolderSubEntries func() error, signatures []int32, skipTargetLookup bool, metadataEvents *[]metadataEvent, pendingChunkDeletes *[]*filer_pb.FileChunk) error {
oldPath, newPath := oldParent.Child(entry.Name()), newParent.Child(newName)
glog.V(1).Infof("moving entry %s => %s", oldPath, newPath)
if oldPath == newPath {
glog.V(1).Infof("skip moving entry %s => %s", oldPath, newPath)
return nil
}
sourceEntry := entry.ShallowClone()
sourceEntry.FullPath = oldPath
var existingTarget *filer.Entry
if !skipTargetLookup {
if targetEntry, findErr := fs.filer.FindEntry(ctx, newPath); findErr == nil {
existingTarget = targetEntry.ShallowClone()
} else if findErr != filer_pb.ErrNotFound {
return findErr
}
}
if existingTarget != nil {
switch {
case existingTarget.IsDirectory() && !entry.IsDirectory():
return fmt.Errorf("%s: %w", existingTarget.FullPath, filer_pb.ErrExistingIsDirectory)
case !existingTarget.IsDirectory() && entry.IsDirectory():
return fmt.Errorf("%s: %w", existingTarget.FullPath, filer_pb.ErrExistingIsFile)
}
if deleteErr := fs.filer.DeleteEntryMetaAndData(
filer.WithSuppressedMetadataEvents(ctx),
newPath,
false,
false,
false,
false,
signatures,
0,
); deleteErr != nil {
return deleteErr
}
}
// add to new directory
newEntry := &filer.Entry{
FullPath: newPath,
Attr: entry.Attr,
Chunks: entry.GetChunks(),
Extended: entry.Extended,
Content: entry.Content,
HardLinkCounter: entry.HardLinkCounter,
HardLinkId: entry.HardLinkId,
Remote: entry.Remote,
Quota: entry.Quota,
}
if skipTargetLookup {
if newEntry.FullPath.IsLongerFileName(fs.filer.MaxFilenameLength) {
return filer_pb.ErrEntryNameTooLong
}
if createErr := fs.filer.Store.InsertEntryKnownAbsent(filer.WithSuppressedMetadataEvents(ctx), newEntry); createErr != nil {
return fmt.Errorf("insert entry %s: %v", newEntry.FullPath, createErr)
}
} else {
if createErr := fs.filer.CreateEntry(filer.WithSuppressedMetadataEvents(ctx), newEntry, nil, false, false, signatures, false, fs.filer.MaxFilenameLength); createErr != nil {
return createErr
}
}
if existingTarget != nil {
toDelete, err := filer.MinusChunks(ctx, fs.filer.MasterClient.GetLookupFileIdFunction(), existingTarget.GetChunks(), newEntry.GetChunks(), fs.filer.MasterClient)
if err != nil {
glog.ErrorfCtx(ctx, "Failed to resolve overwrite target chunks during rename. new: %v, old: %v", newEntry.GetChunks(), existingTarget.GetChunks())
} else if len(toDelete) > 0 {
// Defer chunk deletion until after CommitTransaction so that a
// failure in any subsequent step (child moves, oldPath delete,
// stream send, or the commit itself) leaves the chunks intact for
// the rolled-back rename.
*pendingChunkDeletes = append(*pendingChunkDeletes, toDelete...)
}
}
if stream != nil {
if err := stream.Send(&filer_pb.StreamRenameEntryResponse{
Directory: string(oldParent),
EventNotification: &filer_pb.EventNotification{
OldEntry: &filer_pb.Entry{
Name: entry.Name(),
},
NewEntry: newEntry.ToProtoEntry(),
DeleteChunks: false,
NewParentPath: string(newParent),
IsFromOtherCluster: false,
Signatures: nil,
},
TsNs: time.Now().UnixNano(),
}); err != nil {
return err
}
}
if existingTarget != nil {
*metadataEvents = append(*metadataEvents, metadataEvent{
oldEntry: existingTarget,
deleteChunks: true,
})
}
*metadataEvents = append(*metadataEvents, metadataEvent{
oldEntry: sourceEntry,
newEntry: newEntry,
})
if moveFolderSubEntries != nil {
if moveChildrenErr := moveFolderSubEntries(); moveChildrenErr != nil {
return moveChildrenErr
}
}
// delete old entry
ctx = context.WithValue(ctx, "OP", "MV")
deleteErr := fs.filer.DeleteEntryMetaAndData(filer.WithSuppressedMetadataEvents(ctx), oldPath, false, false, false, false, signatures, 0)
if deleteErr != nil {
return deleteErr
}
if stream != nil {
if err := stream.Send(&filer_pb.StreamRenameEntryResponse{
Directory: string(oldParent),
EventNotification: &filer_pb.EventNotification{
OldEntry: &filer_pb.Entry{
Name: entry.Name(),
},
NewEntry: nil,
DeleteChunks: false,
NewParentPath: "",
IsFromOtherCluster: false,
Signatures: nil,
},
TsNs: time.Now().UnixNano(),
}); err != nil {
return err
}
}
return nil
}