add a new way to manually compact corrupted volume

fix https://github.com/chrislusf/seaweedfs/issues/371

weed/command/compact.go

@@ -23,6 +23,7 @@ var (
 	compactVolumePath       = cmdCompact.Flag.String("dir", ".", "data directory to store files")
 	compactVolumeCollection = cmdCompact.Flag.String("collection", "", "volume collection name")
 	compactVolumeId         = cmdCompact.Flag.Int("volumeId", -1, "a volume id. The volume should already exist in the dir.")
+	compactMethod           = cmdCompact.Flag.Int("method", 0, "option to choose which compact method. use 0 or 1.")
 )
 
 func runCompact(cmd *Command, args []string) bool {
@@ -37,9 +38,15 @@ func runCompact(cmd *Command, args []string) bool {
 	if err != nil {
 		glog.Fatalf("Load Volume [ERROR] %s\n", err)
 	}
+	if *compactMethod == 0 {
 		if err = v.Compact(); err != nil {
 			glog.Fatalf("Compact Volume [ERROR] %s\n", err)
 		}
+	} else {
+		if err = v.Compact2(); err != nil {
+			glog.Fatalf("Compact Volume [ERROR] %s\n", err)
+		}
+	}
 
 	return true
 }

weed/storage/volume_vacuum.go

@@ -23,6 +23,14 @@ func (v *Volume) Compact() error {
 	glog.V(3).Infof("creating copies for volume %d ...", v.Id)
 	return v.copyDataAndGenerateIndexFile(filePath+".cpd", filePath+".cpx")
 }
+
+func (v *Volume) Compact2() error {
+	glog.V(3).Infof("Compact2 ...")
+	filePath := v.FileName()
+	glog.V(3).Infof("creating copies for volume %d ...", v.Id)
+	return v.copyDataBasedOnIndexFile(filePath+".cpd", filePath+".cpx")
+}
+
 func (v *Volume) commitCompact() error {
 	glog.V(3).Infof("Committing vacuuming...")
 	v.dataFileAccessLock.Lock()
@@ -91,3 +99,64 @@ func (v *Volume) copyDataAndGenerateIndexFile(dstName, idxName string) (err erro
 
 	return
 }
+
+func (v *Volume) copyDataBasedOnIndexFile(dstName, idxName string) (err error) {
+	var (
+		dst, idx, oldIndexFile *os.File
+	)
+	if dst, err = os.OpenFile(dstName, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0644); err != nil {
+		return
+	}
+	defer dst.Close()
+
+	if idx, err = os.OpenFile(idxName, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0644); err != nil {
+		return
+	}
+	defer idx.Close()
+
+	if oldIndexFile, err = os.OpenFile(v.FileName()+".idx", os.O_RDONLY, 0644); err != nil {
+		return
+	}
+	defer oldIndexFile.Close()
+
+	nm := NewNeedleMap(idx)
+	now := uint64(time.Now().Unix())
+
+	v.SuperBlock.CompactRevision++
+	dst.Write(v.SuperBlock.Bytes())
+	new_offset := int64(SuperBlockSize)
+
+	WalkIndexFile(oldIndexFile, func(key uint64, offset, size uint32) error {
+		if size <= 0 {
+			return nil
+		}
+
+		nv, ok := v.nm.Get(key)
+		if !ok {
+			return nil
+		}
+
+		n := new(Needle)
+		n.ReadData(v.dataFile, int64(offset)*NeedlePaddingSize, size, v.Version())
+		defer n.ReleaseMemory()
+
+		if n.HasTtl() && now >= n.LastModified+uint64(v.Ttl.Minutes()*60) {
+			return nil
+		}
+
+		glog.V(4).Infoln("needle expected offset ", offset, "ok", ok, "nv", nv)
+		if nv.Offset == offset && nv.Size > 0 {
+			if err = nm.Put(n.Id, uint32(new_offset/NeedlePaddingSize), n.Size); err != nil {
+				return fmt.Errorf("cannot put needle: %s", err)
+			}
+			if _, err = n.Append(dst, v.Version()); err != nil {
+				return fmt.Errorf("cannot append needle: %s", err)
+			}
+			new_offset += n.DiskSize()
+			glog.V(3).Infoln("saving key", n.Id, "volume offset", offset, "=>", new_offset, "data_size", n.Size)
+		}
+		return nil
+	})
+
+	return
+}