seaweedfs/weed/mount/weedfs_file_sync.go
Chris Lu 952afd810c
Fix dead lock (#5815)
* reduce locks to avoid dead lock

Flush->FlushData->uplloadPipeline.FluahAll
uploaderCount>0

goroutine 1 [sync.Cond.Wait, 71 minutes]:
sync.runtime_notifyListWait(0xc0007ae4d0, 0x0)
	/usr/local/go/src/runtime/sema.go:569 +0x159
sync.(*Cond).Wait(0xc001a59290?)
	/usr/local/go/src/sync/cond.go:70 +0x85
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*UploadPipeline).waitForCurrentWritersToComplete(0xc0002ee4d0)
	/github/workspace/weed/mount/page_writer/upload_pipeline_lock.go:58 +0x32
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*UploadPipeline).FlushAll(0xc0002ee4d0)
	/github/workspace/weed/mount/page_writer/upload_pipeline.go:151 +0x25
github.com/seaweedfs/seaweedfs/weed/mount.(*ChunkedDirtyPages).FlushData(0xc00087e840)
	/github/workspace/weed/mount/dirty_pages_chunked.go:54 +0x29
github.com/seaweedfs/seaweedfs/weed/mount.(*PageWriter).FlushData(...)
	/github/workspace/weed/mount/page_writer.go:50
github.com/seaweedfs/seaweedfs/weed/mount.(*WFS).doFlush(0xc0006ad600, 0xc00030d380, 0x0, 0x0)
	/github/workspace/weed/mount/weedfs_file_sync.go:101 +0x169
github.com/seaweedfs/seaweedfs/weed/mount.(*WFS).Flush(0xc0006ad600, 0xc001a594a8?, 0xc0004c1ca0)
	/github/workspace/weed/mount/weedfs_file_sync.go:59 +0x48
github.com/hanwen/go-fuse/v2/fuse.doFlush(0xc0000da870?, 0xc0004c1b08)

SaveContent -> MemChunk.RLock ->
	ChunkedDirtyPages.saveChunkedFileIntervalToStorage
	pages.fh.AddChunks([]*filer_pb.FileChunk{chunk})
		fh.entryLock.Lock()

sync.(*RWMutex).Lock(0x0?)
	/usr/local/go/src/sync/rwmutex.go:146 +0x31
github.com/seaweedfs/seaweedfs/weed/mount.(*FileHandle).AddChunks(0xc00030d380, {0xc00028bdc8, 0x1, 0x1})
	/github/workspace/weed/mount/filehandle.go:93 +0x45
github.com/seaweedfs/seaweedfs/weed/mount.(*ChunkedDirtyPages).saveChunkedFileIntervalToStorage(0xc00087e840, {0x2be7ac0, 0xc00018d9e0}, 0x0, 0x121, 0x17e3c624565ace45, 0x1?)
	/github/workspace/weed/mount/dirty_pages_chunked.go:80 +0x2d4
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*MemChunk).SaveContent(0xc0008d9130, 0xc0008093e0)
	/github/workspace/weed/mount/page_writer/page_chunk_mem.go:115 +0x112
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*UploadPipeline).moveToSealed.func1()
	/github/workspace/weed/mount/page_writer/upload_pipeline.go:187 +0x55
github.com/seaweedfs/seaweedfs/weed/util.(*LimitedConcurrentExecutor).Execute.func1()
	/github/workspace/weed/util/limited_executor.go:38 +0x62
created by github.com/seaweedfs/seaweedfs/weed/util.(*LimitedConcurrentExecutor).Execute in goroutine 1
	/github/workspace/weed/util/limited_executor.go:33 +0x97

On metadata update
fh.entryLock.Lock()
	fh.dirtyPages.Destroy()
		up.chunksLock.Lock => each sealed chunk.FreeReference => MemChunk.Lock

goroutine 134 [sync.RWMutex.Lock, 71 minutes]:
sync.runtime_SemacquireRWMutex(0xc0007c3558?, 0xea?, 0x3fb0800?)
	/usr/local/go/src/runtime/sema.go:87 +0x25
sync.(*RWMutex).Lock(0xc0007c35a8?)
	/usr/local/go/src/sync/rwmutex.go:151 +0x6a
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*MemChunk).FreeResource(0xc0008d9130)
	/github/workspace/weed/mount/page_writer/page_chunk_mem.go:38 +0x2a
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*SealedChunk).FreeReference(0xc00071cdb0, {0xc0006ba1a0, 0x20})
	/github/workspace/weed/mount/page_writer/upload_pipeline.go:38 +0xb7
github.com/seaweedfs/seaweedfs/weed/mount/page_writer.(*UploadPipeline).Shutdown(0xc0002ee4d0)
	/github/workspace/weed/mount/page_writer/upload_pipeline.go:220 +0x185
github.com/seaweedfs/seaweedfs/weed/mount.(*ChunkedDirtyPages).Destroy(0xc0008cea40?)
	/github/workspace/weed/mount/dirty_pages_chunked.go:87 +0x17
github.com/seaweedfs/seaweedfs/weed/mount.(*PageWriter).Destroy(...)
	/github/workspace/weed/mount/page_writer.go:78
github.com/seaweedfs/seaweedfs/weed/mount.NewSeaweedFileSystem.func3({0xc00069a6c0, 0x30}, 0x6?)
	/github/workspace/weed/mount/weedfs.go:119 +0x17a
github.com/seaweedfs/seaweedfs/weed/mount/meta_cache.NewMetaCache.func1({0xc00069a6c0?, 0xc00069a480?}, 0x4015b40?)
	/github/workspace/weed/mount/meta_cache/meta_cache.go:37 +0x1c
github.com/seaweedfs/seaweedfs/weed/mount/meta_cache.SubscribeMetaEvents.func1(0xc000661810)
	/github/workspace/weed/mount/meta_cache/meta_cache_subscribe.go:43 +0x570

* use locked entry everywhere

* modifiable remote entry

* skip locking after getting lock from fhLockTable
2024-07-24 23:46:40 -07:00

187 lines
5.2 KiB
Go

package mount
import (
"context"
"fmt"
"github.com/hanwen/go-fuse/v2/fuse"
"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"
"syscall"
"time"
)
/**
* Flush method
*
* This is called on each close() of the opened file.
*
* Since file descriptors can be duplicated (dup, dup2, fork), for
* one open call there may be many flush calls.
*
* Filesystems shouldn't assume that flush will always be called
* after some writes, or that if will be called at all.
*
* fi->fh will contain the value set by the open method, or will
* be undefined if the open method didn't set any value.
*
* NOTE: the name of the method is misleading, since (unlike
* fsync) the filesystem is not forced to flush pending writes.
* One reason to flush data is if the filesystem wants to return
* write errors during close. However, such use is non-portable
* because POSIX does not require [close] to wait for delayed I/O to
* complete.
*
* If the filesystem supports file locking operations (setlk,
* getlk) it should remove all locks belonging to 'fi->owner'.
*
* If this request is answered with an error code of ENOSYS,
* this is treated as success and future calls to flush() will
* succeed automatically without being send to the filesystem
* process.
*
* Valid replies:
* fuse_reply_err
*
* @param req request handle
* @param ino the inode number
* @param fi file information
*
* [close]: http://pubs.opengroup.org/onlinepubs/9699919799/functions/close.html
*/
func (wfs *WFS) Flush(cancel <-chan struct{}, in *fuse.FlushIn) fuse.Status {
fh := wfs.GetHandle(FileHandleId(in.Fh))
if fh == nil {
return fuse.ENOENT
}
return wfs.doFlush(fh, in.Uid, in.Gid)
}
/**
* Synchronize file contents
*
* If the datasync parameter is non-zero, then only the user data
* should be flushed, not the meta data.
*
* If this request is answered with an error code of ENOSYS,
* this is treated as success and future calls to fsync() will
* succeed automatically without being send to the filesystem
* process.
*
* Valid replies:
* fuse_reply_err
*
* @param req request handle
* @param ino the inode number
* @param datasync flag indicating if only data should be flushed
* @param fi file information
*/
func (wfs *WFS) Fsync(cancel <-chan struct{}, in *fuse.FsyncIn) (code fuse.Status) {
fh := wfs.GetHandle(FileHandleId(in.Fh))
if fh == nil {
return fuse.ENOENT
}
return wfs.doFlush(fh, in.Uid, in.Gid)
}
func (wfs *WFS) doFlush(fh *FileHandle, uid, gid uint32) fuse.Status {
// flush works at fh level
fileFullPath := fh.FullPath()
dir, name := fileFullPath.DirAndName()
// send the data to the OS
glog.V(4).Infof("doFlush %s fh %d", fileFullPath, fh.fh)
if !wfs.IsOverQuota {
if err := fh.dirtyPages.FlushData(); err != nil {
glog.Errorf("%v doFlush: %v", fileFullPath, err)
return fuse.EIO
}
}
if !fh.dirtyMetadata {
return fuse.OK
}
if wfs.IsOverQuota {
return fuse.Status(syscall.ENOSPC)
}
fhActiveLock := fh.wfs.fhLockTable.AcquireLock("doFlush", fh.fh, util.ExclusiveLock)
defer fh.wfs.fhLockTable.ReleaseLock(fh.fh, fhActiveLock)
err := wfs.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
entry := fh.GetEntry()
entry.Name = name // this flush may be just after a rename operation
if entry.Attributes != nil {
entry.Attributes.Mime = fh.contentType
if entry.Attributes.Uid == 0 {
entry.Attributes.Uid = uid
}
if entry.Attributes.Gid == 0 {
entry.Attributes.Gid = gid
}
if entry.Attributes.Crtime == 0 {
entry.Attributes.Crtime = time.Now().Unix()
}
entry.Attributes.Mtime = time.Now().Unix()
}
request := &filer_pb.CreateEntryRequest{
Directory: string(dir),
Entry: entry.GetEntry(),
Signatures: []int32{wfs.signature},
SkipCheckParentDirectory: true,
}
glog.V(4).Infof("%s set chunks: %v", fileFullPath, len(entry.GetChunks()))
//for i, chunk := range entry.GetChunks() {
// glog.V(4).Infof("%s chunks %d: %v [%d,%d)", fileFullPath, i, chunk.GetFileIdString(), chunk.Offset, chunk.Offset+int64(chunk.Size))
//}
manifestChunks, nonManifestChunks := filer.SeparateManifestChunks(entry.GetChunks())
chunks, _ := filer.CompactFileChunks(wfs.LookupFn(), nonManifestChunks)
chunks, manifestErr := filer.MaybeManifestize(wfs.saveDataAsChunk(fileFullPath), chunks)
if manifestErr != nil {
// not good, but should be ok
glog.V(0).Infof("MaybeManifestize: %v", manifestErr)
}
entry.Chunks = append(chunks, manifestChunks...)
wfs.mapPbIdFromLocalToFiler(request.Entry)
defer wfs.mapPbIdFromFilerToLocal(request.Entry)
if err := filer_pb.CreateEntry(client, request); err != nil {
glog.Errorf("fh flush create %s: %v", fileFullPath, err)
return fmt.Errorf("fh flush create %s: %v", fileFullPath, err)
}
wfs.metaCache.InsertEntry(context.Background(), filer.FromPbEntry(request.Directory, request.Entry))
return nil
})
if err == nil {
fh.dirtyMetadata = false
}
if err != nil {
glog.Errorf("%v fh %d flush: %v", fileFullPath, fh.fh, err)
return fuse.EIO
}
if IsDebugFileReadWrite {
fh.mirrorFile.Sync()
}
return fuse.OK
}