2.43

fix https://github.com/chrislusf/seaweedfs/issues/2038

.github/workflows/release.yml

@@ -33,8 +33,9 @@ jobs:
         run: echo BUILD_TIME=$(date -u +%Y-%m-%d-%H-%M) >> ${GITHUB_ENV}
 
       - name: Go Release Binaries
-        uses: wangyoucao577/go-release-action@v1.14
+        uses: wangyoucao577/go-release-action@v1.17
         with:
+          goversion: 1.16
           github_token: ${{ secrets.GITHUB_TOKEN }}
           goos: ${{ matrix.goos }}
           goarch: ${{ matrix.goarch }}
@@ -49,8 +50,9 @@ jobs:
           asset_name: "weed-large-disk-${{ env.BUILD_TIME }}-${{ matrix.goos }}-${{ matrix.goarch }}"
 
       - name: Go Release Binaries
-        uses: wangyoucao577/go-release-action@v1.14
+        uses: wangyoucao577/go-release-action@v1.17
         with:
+          goversion: 1.16
           github_token: ${{ secrets.GITHUB_TOKEN }}
           goos: ${{ matrix.goos }}
           goarch: ${{ matrix.goarch }}

.travis.yml

@@ -1,7 +1,6 @@
 sudo: false
 language: go
 go:
-  - 1.15.x
   - 1.16.x
 
 before_install:

README.md

@@ -68,12 +68,16 @@ Table of Contents
 * [License](#license)
 
 
-## Quick Start ##
+## Quick Start with single binary ##
 * Download the latest binary from https://github.com/chrislusf/seaweedfs/releases and unzip a single binary file `weed` or `weed.exe`
 * Run `weed server -dir=/some/data/dir -s3` to start one master, one volume server, one filer, and one S3 gateway.
 
 Also, to increase capacity, just add more volume servers by running `weed volume -dir="/some/data/dir2" -mserver="<master_host>:9333" -port=8081` locally, or on a different machine, or on thousands of machines. That is it!
 
+## Quick Start for S3 API on Docker ##
+
+`docker run -p 8333:8333 chrislusf/seaweedfs server -s3`
+
 ## Introduction ##
 
 SeaweedFS is a simple and highly scalable distributed file system. There are two objectives:

docker/Makefile

@@ -5,7 +5,9 @@ all: gen
 gen: dev
 
 binary:
-	cd ../weed; CGO_ENABLED=0 GOOS=linux go build -ldflags "-extldflags -static"; mv weed ../docker/
+	export SWCOMMIT=$(shell git rev-parse --short HEAD)
+	export SWLDFLAGS="-X github.com/chrislusf/seaweedfs/weed/util.COMMIT=$(SWCOMMIT)"
+	cd ../weed; CGO_ENABLED=0 GOOS=linux go build -ldflags "-extldflags -static $(SWLDFLAGS)"; mv weed ../docker/
 
 build: binary
 	docker build --no-cache -t chrislusf/seaweedfs:local -f Dockerfile.local .

docker/compose/local-mount-compose.yml

@@ -38,7 +38,7 @@ services:
   mount_2:
     image: chrislusf/seaweedfs:local
     privileged: true
-    entrypoint: '/bin/sh -c "mkdir -p t2 && mkdir -p cache/t2 && weed -v=4  mount -filer=filer:8888 -cacheDir=./cache/t2 -dir=./t2 -filer.path=/c1 -volumeServerAcess=publicUrl"'
+    entrypoint: '/bin/sh -c "mkdir -p t2 && mkdir -p cache/t2 && weed -v=4  mount -filer=filer:8888 -cacheDir=./cache/t2 -dir=./t2 -filer.path=/c1"'
     depends_on:
       - master
       - volume

k8s/seaweedfs/Chart.yaml

@@ -1,5 +1,5 @@
 apiVersion: v1
 description: SeaweedFS
 name: seaweedfs
-appVersion: "2.41"
-version: 2.41
+appVersion: "2.43"
+version: 2.43

k8s/seaweedfs/values.yaml

@@ -4,7 +4,7 @@ global:
   registry: ""
   repository: ""
   imageName: chrislusf/seaweedfs
-  # imageTag: "2.41" - started using {.Chart.appVersion}
+  # imageTag: "2.43" - started using {.Chart.appVersion}
   imagePullPolicy: IfNotPresent
   imagePullSecrets: imagepullsecret
   restartPolicy: Always

unmaintained/repeated_vacuum/repeated_vacuum.go

@@ -52,7 +52,7 @@ func main() {
 }
 
 func genFile(grpcDialOption grpc.DialOption, i int) (*operation.AssignResult, string) {
-	assignResult, err := operation.Assign(*master, grpcDialOption, &operation.VolumeAssignRequest{
+	assignResult, err := operation.Assign(func() string { return *master }, grpcDialOption, &operation.VolumeAssignRequest{
 		Count:       1,
 		Replication: *replication,
 	})

weed/command/server.go

@@ -58,7 +58,8 @@ var (
 	serverDisableHttp         = cmdServer.Flag.Bool("disableHttp", false, "disable http requests, only gRPC operations are allowed.")
 	volumeDataFolders         = cmdServer.Flag.String("dir", os.TempDir(), "directories to store data files. dir[,dir]...")
 	volumeMaxDataVolumeCounts = cmdServer.Flag.String("volume.max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured.")
-	volumeMinFreeSpacePercent = cmdServer.Flag.String("volume.minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly.")
+	volumeMinFreeSpacePercent = cmdServer.Flag.String("volume.minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly (deprecated, use minFreeSpace instead).")
+	volumeMinFreeSpace        = cmdServer.Flag.String("volume.minFreeSpace", "", "min free disk space (value<=100 as percentage like 1, other as human readable bytes, like 10GiB). Low disk space will mark all volumes as ReadOnly.")
 	serverMetricsHttpPort     = cmdServer.Flag.Int("metricsPort", 0, "Prometheus metrics listen port")
 
 	// pulseSeconds              = cmdServer.Flag.Int("pulseSeconds", 5, "number of seconds between heartbeats")
@@ -244,8 +245,8 @@ func runServer(cmd *Command, args []string) bool {
 
 	// start volume server
 	if *isStartingVolumeServer {
-		go serverOptions.v.startVolumeServer(*volumeDataFolders, *volumeMaxDataVolumeCounts, *serverWhiteListOption, *volumeMinFreeSpacePercent)
-
+		minFreeSpaces := util.MustParseMinFreeSpace(*minFreeSpace, *minFreeSpacePercent)
+		go serverOptions.v.startVolumeServer(*volumeDataFolders, *volumeMaxDataVolumeCounts, *serverWhiteListOption, minFreeSpaces)
 	}
 
 	if *isStartingMasterServer {
@@ -253,6 +254,4 @@ func runServer(cmd *Command, args []string) bool {
 	}
 
 	select {}
-
-	return true
 }

weed/command/volume.go

@@ -57,7 +57,6 @@ type VolumeServerOptions struct {
 	compactionMBPerSecond   *int
 	fileSizeLimitMB         *int
 	concurrentUploadLimitMB *int
-	minFreeSpacePercents    []float32
 	pprof                   *bool
 	preStopSeconds          *int
 	metricsHttpPort         *int
@@ -105,7 +104,8 @@ var (
 	volumeFolders         = cmdVolume.Flag.String("dir", os.TempDir(), "directories to store data files. dir[,dir]...")
 	maxVolumeCounts       = cmdVolume.Flag.String("max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured.")
 	volumeWhiteListOption = cmdVolume.Flag.String("whiteList", "", "comma separated Ip addresses having write permission. No limit if empty.")
-	minFreeSpacePercent   = cmdVolume.Flag.String("minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly.")
+	minFreeSpacePercent   = cmdVolume.Flag.String("minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly (deprecated, use minFreeSpace instead).")
+	minFreeSpace          = cmdVolume.Flag.String("minFreeSpace", "", "min free disk space (value<=100 as percentage like 1, other as human readable bytes, like 10GiB). Low disk space will mark all volumes as ReadOnly.")
 )
 
 func runVolume(cmd *Command, args []string) bool {
@@ -120,12 +120,13 @@ func runVolume(cmd *Command, args []string) bool {
 
 	go stats_collect.StartMetricsServer(*v.metricsHttpPort)
 
-	v.startVolumeServer(*volumeFolders, *maxVolumeCounts, *volumeWhiteListOption, *minFreeSpacePercent)
+	minFreeSpaces := util.MustParseMinFreeSpace(*minFreeSpace, *minFreeSpacePercent)
+	v.startVolumeServer(*volumeFolders, *maxVolumeCounts, *volumeWhiteListOption, minFreeSpaces)
 
 	return true
 }
 
-func (v VolumeServerOptions) startVolumeServer(volumeFolders, maxVolumeCounts, volumeWhiteListOption, minFreeSpacePercent string) {
+func (v VolumeServerOptions) startVolumeServer(volumeFolders, maxVolumeCounts, volumeWhiteListOption string, minFreeSpaces []util.MinFreeSpace) {
 
 	// Set multiple folders and each folder's max volume count limit'
 	v.folders = strings.Split(volumeFolders, ",")
@@ -153,22 +154,13 @@ func (v VolumeServerOptions) startVolumeServer(volumeFolders, maxVolumeCounts, v
 		glog.Fatalf("%d directories by -dir, but only %d max is set by -max", len(v.folders), len(v.folderMaxLimits))
 	}
 
-	// set minFreeSpacePercent
-	minFreeSpacePercentStrings := strings.Split(minFreeSpacePercent, ",")
-	for _, freeString := range minFreeSpacePercentStrings {
-		if value, e := strconv.ParseFloat(freeString, 32); e == nil {
-			v.minFreeSpacePercents = append(v.minFreeSpacePercents, float32(value))
-		} else {
-			glog.Fatalf("The value specified in -minFreeSpacePercent not a valid value %s", freeString)
-		}
-	}
-	if len(v.minFreeSpacePercents) == 1 && len(v.folders) > 1 {
+	if len(minFreeSpaces) == 1 && len(v.folders) > 1 {
 		for i := 0; i < len(v.folders)-1; i++ {
-			v.minFreeSpacePercents = append(v.minFreeSpacePercents, v.minFreeSpacePercents[0])
+			minFreeSpaces = append(minFreeSpaces, minFreeSpaces[0])
 		}
 	}
-	if len(v.folders) != len(v.minFreeSpacePercents) {
-		glog.Fatalf("%d directories by -dir, but only %d minFreeSpacePercent is set by -minFreeSpacePercent", len(v.folders), len(v.minFreeSpacePercents))
+	if len(v.folders) != len(minFreeSpaces) {
+		glog.Fatalf("%d directories by -dir, but only %d minFreeSpacePercent is set by -minFreeSpacePercent", len(v.folders), len(minFreeSpaces))
 	}
 
 	// set disk types
@@ -231,7 +223,7 @@ func (v VolumeServerOptions) startVolumeServer(volumeFolders, maxVolumeCounts, v
 
 	volumeServer := weed_server.NewVolumeServer(volumeMux, publicVolumeMux,
 		*v.ip, *v.port, *v.publicUrl,
-		v.folders, v.folderMaxLimits, v.minFreeSpacePercents, diskTypes,
+		v.folders, v.folderMaxLimits, minFreeSpaces, diskTypes,
 		*v.idxFolder,
 		volumeNeedleMapKind,
 		strings.Split(masters, ","), 5, *v.dataCenter, *v.rack,

weed/filer/filechunks.go

@@ -2,7 +2,6 @@ package filer
 
 import (
 	"bytes"
-	"encoding/hex"
 	"fmt"
 	"github.com/chrislusf/seaweedfs/weed/wdclient"
 	"math"
@@ -43,12 +42,11 @@ func ETagEntry(entry *Entry) (etag string) {
 
 func ETagChunks(chunks []*filer_pb.FileChunk) (etag string) {
 	if len(chunks) == 1 {
-		return chunks[0].ETag
+		return fmt.Sprintf("%x", util.Base64Md5ToBytes(chunks[0].ETag))
 	}
 	md5_digests := [][]byte{}
 	for _, c := range chunks {
-		md5_decoded, _ := hex.DecodeString(c.ETag)
-		md5_digests = append(md5_digests, md5_decoded)
+		md5_digests = append(md5_digests, util.Base64Md5ToBytes(c.ETag))
 	}
 	return fmt.Sprintf("%x-%d", util.Md5(bytes.Join(md5_digests, nil)), len(chunks))
 }

weed/filer/reader_at.go

@@ -139,13 +139,15 @@ func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, err error) {
 		}
 		glog.V(4).Infof("read [%d,%d), %d/%d chunk %s [%d,%d)", chunkStart, chunkStop, i, len(c.chunkViews), chunk.FileId, chunk.LogicOffset-chunk.Offset, chunk.LogicOffset-chunk.Offset+int64(chunk.Size))
 		var buffer []byte
-		buffer, err = c.readFromWholeChunkData(chunk, nextChunk)
+		bufferOffset := chunkStart - chunk.LogicOffset + chunk.Offset
+		bufferLength := chunkStop - chunkStart
+		buffer, err = c.readChunkSlice(chunk, nextChunk, uint64(bufferOffset), uint64(bufferLength))
 		if err != nil {
 			glog.Errorf("fetching chunk %+v: %v\n", chunk, err)
 			return
 		}
-		bufferOffset := chunkStart - chunk.LogicOffset + chunk.Offset
-		copied := copy(p[startOffset-offset:chunkStop-chunkStart+startOffset-offset], buffer[bufferOffset:bufferOffset+chunkStop-chunkStart])
+
+		copied := copy(p[startOffset-offset:chunkStop-chunkStart+startOffset-offset], buffer)
 		n += copied
 		startOffset, remaining = startOffset+int64(copied), remaining-int64(copied)
 	}
@@ -167,6 +169,20 @@ func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, err error) {
 
 }
 
+func (c *ChunkReadAt) readChunkSlice(chunkView *ChunkView, nextChunkViews *ChunkView, offset, length uint64) ([]byte, error) {
+
+	chunkSlice := c.chunkCache.GetChunkSlice(chunkView.FileId, offset, length)
+	if len(chunkSlice) > 0 {
+		return chunkSlice, nil
+	}
+	chunkData, err := c.readFromWholeChunkData(chunkView, nextChunkViews)
+	if err != nil {
+		return nil, err
+	}
+	wanted := min(int64(length), int64(len(chunkData))-int64(offset))
+	return chunkData[offset : int64(offset)+wanted], nil
+}
+
 func (c *ChunkReadAt) readFromWholeChunkData(chunkView *ChunkView, nextChunkViews ...*ChunkView) (chunkData []byte, err error) {
 
 	if c.lastChunkFileId == chunkView.FileId {

weed/filer/reader_at_test.go

@@ -20,6 +20,11 @@ func (m *mockChunkCache) GetChunk(fileId string, minSize uint64) (data []byte) {
 	}
 	return data
 }
+
+func(m *mockChunkCache) GetChunkSlice(fileId string, offset, length uint64) []byte {
+	return nil
+}
+
 func (m *mockChunkCache) SetChunk(fileId string, data []byte) {
 }
 

weed/filesys/dir.go

@@ -29,7 +29,7 @@ type Dir struct {
 
 var _ = fs.Node(&Dir{})
 
-//var _ = fs.NodeIdentifier(&Dir{})
+var _ = fs.NodeIdentifier(&Dir{})
 var _ = fs.NodeCreater(&Dir{})
 var _ = fs.NodeMknoder(&Dir{})
 var _ = fs.NodeMkdirer(&Dir{})
@@ -45,7 +45,10 @@ var _ = fs.NodeRemovexattrer(&Dir{})
 var _ = fs.NodeListxattrer(&Dir{})
 var _ = fs.NodeForgetter(&Dir{})
 
-func (dir *Dir) xId() uint64 {
+func (dir *Dir) Id() uint64 {
+	if dir.parent == nil {
+		return 1
+	}
 	return dir.id
 }
 
@@ -66,7 +69,7 @@ func (dir *Dir) Attr(ctx context.Context, attr *fuse.Attr) error {
 		return err
 	}
 
-	// attr.Inode = dir.Id()
+	attr.Inode = dir.Id()
 	attr.Mode = os.FileMode(entry.Attributes.FileMode) | os.ModeDir
 	attr.Mtime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Crtime = time.Unix(entry.Attributes.Crtime, 0)
@@ -93,7 +96,7 @@ func (dir *Dir) Getxattr(ctx context.Context, req *fuse.GetxattrRequest, resp *f
 func (dir *Dir) setRootDirAttributes(attr *fuse.Attr) {
 	// attr.Inode = 1 // filer2.FullPath(dir.Path).AsInode()
 	attr.Valid = time.Second
-	attr.Inode = 1 // dir.Id()
+	attr.Inode = dir.Id()
 	attr.Uid = dir.wfs.option.MountUid
 	attr.Gid = dir.wfs.option.MountGid
 	attr.Mode = dir.wfs.option.MountMode
@@ -328,6 +331,7 @@ func (dir *Dir) Lookup(ctx context.Context, req *fuse.LookupRequest, resp *fuse.
 		// resp.EntryValid = time.Second
 		resp.Attr.Inode = fullFilePath.AsInode()
 		resp.Attr.Valid = time.Second
+		resp.Attr.Size = localEntry.FileSize
 		resp.Attr.Mtime = localEntry.Attr.Mtime
 		resp.Attr.Crtime = localEntry.Attr.Crtime
 		resp.Attr.Mode = localEntry.Attr.Mode

weed/filesys/meta_cache/meta_cache.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"fmt"
 	"os"
-	"strings"
 	"sync"
 
 	"github.com/chrislusf/seaweedfs/weed/filer"
@@ -31,9 +30,6 @@ func NewMetaCache(dbFolder string, baseDir util.FullPath, uidGidMapper *UidGidMa
 		visitedBoundary: bounded_tree.NewBoundedTree(baseDir),
 		uidGidMapper:    uidGidMapper,
 		invalidateFunc: func(fullpath util.FullPath) {
-			if baseDir != "/" && strings.HasPrefix(string(fullpath), string(baseDir)) {
-				fullpath = fullpath[len(baseDir):]
-			}
 			invalidateFunc(fullpath)
 		},
 	}

weed/filesys/wfs.go

@@ -111,6 +111,9 @@ func NewSeaweedFileSystem(option *Option) *WFS {
 
 		dir, name := filePath.DirAndName()
 		parent := NodeWithId(util.FullPath(dir).AsInode())
+		if dir == option.FilerMountRootPath {
+			parent = NodeWithId(1)
+		}
 		if err := wfs.Server.InvalidateEntry(parent, name); err != nil {
 			glog.V(4).Infof("InvalidateEntry %s : %v", filePath, err)
 		}
@@ -121,7 +124,7 @@ func NewSeaweedFileSystem(option *Option) *WFS {
 		wfs.metaCache.Shutdown()
 	})
 
-	wfs.root = &Dir{name: wfs.option.FilerMountRootPath, wfs: wfs}
+	wfs.root = &Dir{name: wfs.option.FilerMountRootPath, wfs: wfs, id: 1}
 	wfs.fsNodeCache = newFsCache(wfs.root)
 
 	if wfs.option.ConcurrentWriters > 0 {

weed/iamapi/iamapi_management_handlers.go

@@ -114,7 +114,11 @@ func (iama *IamApiServer) ListUsers(s3cfg *iam_pb.S3ApiConfiguration, values url
 
 func (iama *IamApiServer) ListAccessKeys(s3cfg *iam_pb.S3ApiConfiguration, values url.Values) (resp ListAccessKeysResponse) {
 	status := iam.StatusTypeActive
+	userName := values.Get("UserName")
 	for _, ident := range s3cfg.Identities {
+		if userName != "" && userName != ident.Name {
+			continue
+		}
 		for _, cred := range ident.Credentials {
 			resp.ListAccessKeysResult.AccessKeyMetadata = append(resp.ListAccessKeysResult.AccessKeyMetadata,
 				&iam.AccessKeyMetadata{UserName: &ident.Name, AccessKeyId: &cred.AccessKey, Status: &status},

weed/operation/upload_content.go

@@ -39,7 +39,7 @@ func (uploadResult *UploadResult) ToPbFileChunk(fileId string, offset int64) *fi
 		Offset:       offset,
 		Size:         uint64(uploadResult.Size),
 		Mtime:        time.Now().UnixNano(),
-		ETag:         uploadResult.ETag,
+		ETag:         uploadResult.ContentMd5,
 		CipherKey:    uploadResult.CipherKey,
 		IsCompressed: uploadResult.Gzip > 0,
 		Fid:          fid,

weed/s3api/auth_credentials.go

@@ -3,14 +3,14 @@ package s3api
 import (
 	"fmt"
 	"github.com/chrislusf/seaweedfs/weed/filer"
-	"github.com/chrislusf/seaweedfs/weed/s3api/s3_constants"
-	"io/ioutil"
-	"net/http"
-
 	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/chrislusf/seaweedfs/weed/pb/iam_pb"
 	xhttp "github.com/chrislusf/seaweedfs/weed/s3api/http"
+	"github.com/chrislusf/seaweedfs/weed/s3api/s3_constants"
 	"github.com/chrislusf/seaweedfs/weed/s3api/s3err"
+	"io/ioutil"
+	"net/http"
+	"strings"
 )
 
 type Action string
@@ -255,11 +255,21 @@ func (identity *Identity) canDo(action Action, bucket string) bool {
 	limitedByBucket := string(action) + ":" + bucket
 	adminLimitedByBucket := s3_constants.ACTION_ADMIN + ":" + bucket
 	for _, a := range identity.Actions {
-		if string(a) == limitedByBucket {
-			return true
-		}
-		if string(a) == adminLimitedByBucket {
-			return true
+		act := string(a)
+		if strings.HasSuffix(act, "*") {
+			if strings.HasPrefix(limitedByBucket, act[:len(act)-1]) {
+				return true
+			}
+			if strings.HasPrefix(adminLimitedByBucket, act[:len(act)-1]) {
+				return true
+			}
+		} else {
+			if act == limitedByBucket {
+				return true
+			}
+			if act == adminLimitedByBucket {
+				return true
+			}
 		}
 	}
 	return false

weed/server/filer_server_handlers_write_upload.go

@@ -1,6 +1,7 @@
 package weed_server
 
 import (
+	"bytes"
 	"crypto/md5"
 	"hash"
 	"io"
@@ -71,6 +72,11 @@ func (fs *FilerServer) uploadReaderToChunks(w http.ResponseWriter, r *http.Reque
 		if uploadResult.Size == 0 {
 			break
 		}
+		uploadedMd5 := util.Base64Md5ToBytes(uploadResult.ContentMd5)
+		readedMd5 := md5Hash.Sum(nil)
+		if !bytes.Equal(uploadedMd5, readedMd5) {
+			glog.Errorf("md5 %x does not match %x uploaded chunk %s to the volume server", readedMd5, uploadedMd5, uploadResult.Name)
+		}
 
 		// Save to chunk manifest structure
 		fileChunks = append(fileChunks, uploadResult.ToPbFileChunk(fileId, chunkOffset))

weed/server/volume_server.go

@@ -43,7 +43,7 @@ type VolumeServer struct {
 
 func NewVolumeServer(adminMux, publicMux *http.ServeMux, ip string,
 	port int, publicUrl string,
-	folders []string, maxCounts []int, minFreeSpacePercents []float32, diskTypes []types.DiskType,
+	folders []string, maxCounts []int, minFreeSpaces []util.MinFreeSpace, diskTypes []types.DiskType,
 	idxFolder string,
 	needleMapKind storage.NeedleMapKind,
 	masterNodes []string, pulseSeconds int,
@@ -85,7 +85,7 @@ func NewVolumeServer(adminMux, publicMux *http.ServeMux, ip string,
 
 	vs.checkWithMaster()
 
-	vs.store = storage.NewStore(vs.grpcDialOption, port, ip, publicUrl, folders, maxCounts, minFreeSpacePercents, idxFolder, vs.needleMapKind, diskTypes)
+	vs.store = storage.NewStore(vs.grpcDialOption, port, ip, publicUrl, folders, maxCounts, minFreeSpaces, idxFolder, vs.needleMapKind, diskTypes)
 	vs.guard = security.NewGuard(whiteList, signingKey, expiresAfterSec, readSigningKey, readExpiresAfterSec)
 
 	handleStaticResources(adminMux)

weed/storage/disk_location.go

@@ -23,7 +23,7 @@ type DiskLocation struct {
 	DiskType               types.DiskType
 	MaxVolumeCount         int
 	OriginalMaxVolumeCount int
-	MinFreeSpacePercent    float32
+	MinFreeSpace           util.MinFreeSpace
 	volumes                map[needle.VolumeId]*Volume
 	volumesLock            sync.RWMutex
 
@@ -34,7 +34,7 @@ type DiskLocation struct {
 	isDiskSpaceLow bool
 }
 
-func NewDiskLocation(dir string, maxVolumeCount int, minFreeSpacePercent float32, idxDir string, diskType types.DiskType) *DiskLocation {
+func NewDiskLocation(dir string, maxVolumeCount int, minFreeSpace util.MinFreeSpace, idxDir string, diskType types.DiskType) *DiskLocation {
 	dir = util.ResolvePath(dir)
 	if idxDir == "" {
 		idxDir = dir
@@ -47,7 +47,7 @@ func NewDiskLocation(dir string, maxVolumeCount int, minFreeSpacePercent float32
 		DiskType:               diskType,
 		MaxVolumeCount:         maxVolumeCount,
 		OriginalMaxVolumeCount: maxVolumeCount,
-		MinFreeSpacePercent:    minFreeSpacePercent,
+		MinFreeSpace:           minFreeSpace,
 	}
 	location.volumes = make(map[needle.VolumeId]*Volume)
 	location.ecVolumes = make(map[needle.VolumeId]*erasure_coding.EcVolume)
@@ -361,14 +361,18 @@ func (l *DiskLocation) CheckDiskSpace() {
 			stats.VolumeServerResourceGauge.WithLabelValues(l.Directory, "all").Set(float64(s.All))
 			stats.VolumeServerResourceGauge.WithLabelValues(l.Directory, "used").Set(float64(s.Used))
 			stats.VolumeServerResourceGauge.WithLabelValues(l.Directory, "free").Set(float64(s.Free))
-			if (s.PercentFree < l.MinFreeSpacePercent) != l.isDiskSpaceLow {
+
+			isLow, desc := l.MinFreeSpace.IsLow(s.Free, s.PercentFree)
+			if isLow != l.isDiskSpaceLow {
 				l.isDiskSpaceLow = !l.isDiskSpaceLow
 			}
+
+			logLevel := glog.Level(4)
 			if l.isDiskSpaceLow {
-				glog.V(0).Infof("dir %s freePercent %.2f%% < min %.2f%%, isLowDiskSpace: %v", dir, s.PercentFree, l.MinFreeSpacePercent, l.isDiskSpaceLow)
-			} else {
-				glog.V(4).Infof("dir %s freePercent %.2f%% < min %.2f%%, isLowDiskSpace: %v", dir, s.PercentFree, l.MinFreeSpacePercent, l.isDiskSpaceLow)
+				logLevel = glog.Level(0)
 			}
+
+			glog.V(logLevel).Infof("dir %s %s", dir, desc)
 		}
 		time.Sleep(time.Minute)
 	}

weed/storage/store.go

@@ -2,6 +2,7 @@ package storage
 
 import (
 	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/util"
 	"path/filepath"
 	"strings"
 	"sync/atomic"
@@ -52,11 +53,12 @@ func (s *Store) String() (str string) {
 	return
 }
 
-func NewStore(grpcDialOption grpc.DialOption, port int, ip, publicUrl string, dirnames []string, maxVolumeCounts []int, minFreeSpacePercents []float32, idxFolder string, needleMapKind NeedleMapKind, diskTypes []DiskType) (s *Store) {
+func NewStore(grpcDialOption grpc.DialOption, port int, ip, publicUrl string, dirnames []string, maxVolumeCounts []int,
+	minFreeSpaces []util.MinFreeSpace, idxFolder string, needleMapKind NeedleMapKind, diskTypes []DiskType) (s *Store) {
 	s = &Store{grpcDialOption: grpcDialOption, Port: port, Ip: ip, PublicUrl: publicUrl, NeedleMapKind: needleMapKind}
 	s.Locations = make([]*DiskLocation, 0)
 	for i := 0; i < len(dirnames); i++ {
-		location := NewDiskLocation(dirnames[i], maxVolumeCounts[i], minFreeSpacePercents[i], idxFolder, diskTypes[i])
+		location := NewDiskLocation(dirnames[i], maxVolumeCounts[i], minFreeSpaces[i], idxFolder, diskTypes[i])
 		location.loadExistingVolumes(needleMapKind)
 		s.Locations = append(s.Locations, location)
 		stats.VolumeServerMaxVolumeCounter.Add(float64(maxVolumeCounts[i]))

weed/util/bounded_tree/bounded_tree.go

@@ -41,9 +41,6 @@ func (t *BoundedTree) EnsureVisited(p util.FullPath, visitFn VisitNodeFunc) (vis
 	if t.root == nil {
 		return
 	}
-	if t.baseDir != "/" {
-		p = p[len(t.baseDir):]
-	}
 	components := p.Split()
 	// fmt.Printf("components %v %d\n", components, len(components))
 	canDelete, err := t.ensureVisited(t.root, t.baseDir, components, 0, visitFn)

weed/util/bytes.go

@@ -7,6 +7,10 @@ import (
 	"encoding/base64"
 	"fmt"
 	"io"
+	"math"
+	"strconv"
+	"strings"
+	"unicode"
 )
 
 // BytesToHumanReadable returns the converted human readable representation of the bytes.
@@ -161,3 +165,94 @@ func NewBytesReader(b []byte) *BytesReader {
 		Reader: bytes.NewReader(b),
 	}
 }
+
+// EmptyTo returns to if s is empty.
+func EmptyTo(s, to string) string {
+	if s == "" {
+		return to
+	}
+
+	return s
+}
+
+// IfElse works like b ? this : that.
+func IfElse(b bool, this, that string) string {
+	if b {
+		return this
+	}
+	return that
+}
+
+// ParseBytes parses a string representation of bytes into the number
+// of bytes it represents.
+//
+// See Also: Bytes, IBytes.
+//
+// ParseBytes("42MB") -> 42000000, nil
+// ParseBytes("42 MB") -> 42000000, nil
+// ParseBytes("42 mib") -> 44040192, nil
+func ParseBytes(s string) (uint64, error) {
+	lastDigit := 0
+	hasComma := false
+	for _, r := range s {
+		if !(unicode.IsDigit(r) || r == '.' || r == ',') {
+			break
+		}
+		if r == ',' {
+			hasComma = true
+		}
+		lastDigit++
+	}
+
+	num := s[:lastDigit]
+	if hasComma {
+		num = strings.Replace(num, ",", "", -1)
+	}
+
+	f, err := strconv.ParseFloat(num, 64)
+	if err != nil {
+		return 0, err
+	}
+
+	extra := strings.ToLower(strings.TrimSpace(s[lastDigit:]))
+	if m, ok := bytesSizeTable[extra]; ok {
+		f *= float64(m)
+		if f >= math.MaxUint64 {
+			return 0, fmt.Errorf("too large: %v", s)
+		}
+		return uint64(f), nil
+	}
+
+	return 0, fmt.Errorf("unhandled size name: %v", extra)
+}
+
+var bytesSizeTable = map[string]uint64{
+	"b": Byte, "kib": KiByte, "kb": KByte, "mib": MiByte, "mb": MByte, "gib": GiByte, "gb": GByte,
+	"tib": TiByte, "tb": TByte, "pib": PiByte, "pb": PByte, "eib": EiByte, "eb": EByte,
+	// Without suffix
+	"": Byte, "ki": KiByte, "k": KByte, "mi": MiByte, "m": MByte, "gi": GiByte, "g": GByte,
+	"ti": TiByte, "t": TByte, "pi": PiByte, "p": PByte, "ei": EiByte, "e": EByte,
+}
+
+// IEC Sizes.
+// kibis of bits
+const (
+	Byte = 1 << (iota * 10)
+	KiByte
+	MiByte
+	GiByte
+	TiByte
+	PiByte
+	EiByte
+)
+
+// SI Sizes.
+const (
+	IByte = 1
+	KByte = IByte * 1000
+	MByte = KByte * 1000
+	GByte = MByte * 1000
+	TByte = GByte * 1000
+	PByte = TByte * 1000
+	EByte = PByte * 1000
+)

weed/util/bytes_test.go

@@ -0,0 +1,59 @@
+package util
+
+import "testing"
+
+func TestByteParsing(t *testing.T) {
+	tests := []struct {
+		in  string
+		exp uint64
+	}{
+		{"42", 42},
+		{"42MB", 42000000},
+		{"42MiB", 44040192},
+		{"42mb", 42000000},
+		{"42mib", 44040192},
+		{"42MIB", 44040192},
+		{"42 MB", 42000000},
+		{"42 MiB", 44040192},
+		{"42 mb", 42000000},
+		{"42 mib", 44040192},
+		{"42 MIB", 44040192},
+		{"42.5MB", 42500000},
+		{"42.5MiB", 44564480},
+		{"42.5 MB", 42500000},
+		{"42.5 MiB", 44564480},
+		// No need to say B
+		{"42M", 42000000},
+		{"42Mi", 44040192},
+		{"42m", 42000000},
+		{"42mi", 44040192},
+		{"42MI", 44040192},
+		{"42 M", 42000000},
+		{"42 Mi", 44040192},
+		{"42 m", 42000000},
+		{"42 mi", 44040192},
+		{"42 MI", 44040192},
+		{"42.5M", 42500000},
+		{"42.5Mi", 44564480},
+		{"42.5 M", 42500000},
+		{"42.5 Mi", 44564480},
+		// Bug #42
+		{"1,005.03 MB", 1005030000},
+		// Large testing, breaks when too much larger than
+		// this.
+		{"12.5 EB", uint64(12.5 * float64(EByte))},
+		{"12.5 E", uint64(12.5 * float64(EByte))},
+		{"12.5 EiB", uint64(12.5 * float64(EiByte))},
+	}
+
+	for _, p := range tests {
+		got, err := ParseBytes(p.in)
+		if err != nil {
+			t.Errorf("Couldn't parse %v: %v", p.in, err)
+		}
+		if got != p.exp {
+			t.Errorf("Expected %v for %v, got %v",
+				p.exp, p.in, got)
+		}
+	}
+}

weed/util/chunk_cache/chunk_cache.go

@@ -1,14 +1,18 @@
 package chunk_cache
 
 import (
+	"errors"
 	"sync"
 
 	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/chrislusf/seaweedfs/weed/storage/needle"
 )
 
+var ErrorOutOfBounds = errors.New("attempt to read out of bounds")
+
 type ChunkCache interface {
 	GetChunk(fileId string, minSize uint64) (data []byte)
+	GetChunkSlice(fileId string, offset, length uint64) []byte
 	SetChunk(fileId string, data []byte)
 }
 
@@ -22,6 +26,8 @@ type TieredChunkCache struct {
 	onDiskCacheSizeLimit2 uint64
 }
 
+var _ ChunkCache = &TieredChunkCache{}
+
 func NewTieredChunkCache(maxEntries int64, dir string, diskSizeInUnit int64, unitSize int64) *TieredChunkCache {
 
 	c := &TieredChunkCache{
@@ -87,6 +93,58 @@ func (c *TieredChunkCache) doGetChunk(fileId string, minSize uint64) (data []byt
 
 }
 
+func (c *TieredChunkCache) GetChunkSlice(fileId string, offset, length uint64) []byte {
+	if c == nil {
+		return nil
+	}
+
+	c.RLock()
+	defer c.RUnlock()
+
+	return c.doGetChunkSlice(fileId, offset, length)
+}
+
+func (c *TieredChunkCache) doGetChunkSlice(fileId string, offset, length uint64) (data []byte) {
+
+	minSize := offset + length
+	if minSize <= c.onDiskCacheSizeLimit0 {
+		data, err := c.memCache.getChunkSlice(fileId, offset, length)
+		if err != nil {
+			glog.Errorf("failed to read from memcache: %s", err)
+		}
+		if len(data) >= int(minSize) {
+			return data
+		}
+	}
+
+	fid, err := needle.ParseFileIdFromString(fileId)
+	if err != nil {
+		glog.Errorf("failed to parse file id %s", fileId)
+		return nil
+	}
+
+	if minSize <= c.onDiskCacheSizeLimit0 {
+		data = c.diskCaches[0].getChunkSlice(fid.Key, offset, length)
+		if len(data) >= int(minSize) {
+			return data
+		}
+	}
+	if minSize <= c.onDiskCacheSizeLimit1 {
+		data = c.diskCaches[1].getChunkSlice(fid.Key, offset, length)
+		if len(data) >= int(minSize) {
+			return data
+		}
+	}
+	{
+		data = c.diskCaches[2].getChunkSlice(fid.Key, offset, length)
+		if len(data) >= int(minSize) {
+			return data
+		}
+	}
+
+	return nil
+}
+
 func (c *TieredChunkCache) SetChunk(fileId string, data []byte) {
 	if c == nil {
 		return
@@ -131,3 +189,10 @@ func (c *TieredChunkCache) Shutdown() {
 		diskCache.shutdown()
 	}
 }
+
+func min(x, y int) int {
+	if x < y {
+		return x
+	}
+	return y
+}

weed/util/chunk_cache/chunk_cache_in_memory.go

@@ -31,6 +31,20 @@ func (c *ChunkCacheInMemory) GetChunk(fileId string) []byte {
 	return data
 }
 
+func (c *ChunkCacheInMemory) getChunkSlice(fileId string, offset, length uint64) ([]byte, error) {
+	item := c.cache.Get(fileId)
+	if item == nil {
+		return nil, nil
+	}
+	data := item.Value().([]byte)
+	item.Extend(time.Hour)
+	wanted := min(int(length), len(data)-int(offset))
+	if wanted < 0 {
+		return nil, ErrorOutOfBounds
+	}
+	return data[offset : int(offset)+wanted], nil
+}
+
 func (c *ChunkCacheInMemory) SetChunk(fileId string, data []byte) {
 	localCopy := make([]byte, len(data))
 	copy(localCopy, data)

weed/util/chunk_cache/chunk_cache_on_disk.go

@@ -90,11 +90,11 @@ func (v *ChunkCacheVolume) Shutdown() {
 
 func (v *ChunkCacheVolume) doReset() {
 	v.Shutdown()
-	os.Truncate(v.fileName + ".dat", 0)
-	os.Truncate(v.fileName + ".idx", 0)
-	glog.V(4).Infof("cache removeAll %s ...", v.fileName + ".ldb")
+	os.Truncate(v.fileName+".dat", 0)
+	os.Truncate(v.fileName+".idx", 0)
+	glog.V(4).Infof("cache removeAll %s ...", v.fileName+".ldb")
 	os.RemoveAll(v.fileName + ".ldb")
-	glog.V(4).Infof("cache removed %s", v.fileName + ".ldb")
+	glog.V(4).Infof("cache removed %s", v.fileName+".ldb")
 }
 
 func (v *ChunkCacheVolume) Reset() (*ChunkCacheVolume, error) {
@@ -121,6 +121,29 @@ func (v *ChunkCacheVolume) GetNeedle(key types.NeedleId) ([]byte, error) {
 	return data, nil
 }
 
+func (v *ChunkCacheVolume) getNeedleSlice(key types.NeedleId, offset, length uint64) ([]byte, error) {
+	nv, ok := v.nm.Get(key)
+	if !ok {
+		return nil, storage.ErrorNotFound
+	}
+	wanted := min(int(length), int(nv.Size)-int(offset))
+	if wanted < 0 {
+		// should never happen, but better than panicing
+		return nil, ErrorOutOfBounds
+	}
+	data := make([]byte, wanted)
+	if readSize, readErr := v.DataBackend.ReadAt(data, nv.Offset.ToActualOffset()+int64(offset)); readErr != nil {
+		return nil, fmt.Errorf("read %s.dat [%d,%d): %v",
+			v.fileName, nv.Offset.ToActualOffset()+int64(offset), int(nv.Offset.ToActualOffset())+int(offset)+wanted, readErr)
+	} else {
+		if readSize != wanted {
+			return nil, fmt.Errorf("read %d, expected %d", readSize, wanted)
+		}
+	}
+
+	return data, nil
+}
+
 func (v *ChunkCacheVolume) WriteNeedle(key types.NeedleId, data []byte) error {
 
 	offset := v.fileSize

weed/util/chunk_cache/on_disk_cache_layer.go

@@ -82,6 +82,28 @@ func (c *OnDiskCacheLayer) getChunk(needleId types.NeedleId) (data []byte) {
 
 }
 
+func (c *OnDiskCacheLayer) getChunkSlice(needleId types.NeedleId, offset, length uint64) (data []byte) {
+
+	var err error
+
+	for _, diskCache := range c.diskCaches {
+		data, err = diskCache.getNeedleSlice(needleId, offset, length)
+		if err == storage.ErrorNotFound {
+			continue
+		}
+		if err != nil {
+			glog.Errorf("failed to read cache file %s id %d", diskCache.fileName, needleId)
+			continue
+		}
+		if len(data) != 0 {
+			return
+		}
+	}
+
+	return nil
+
+}
+
 func (c *OnDiskCacheLayer) shutdown() {
 
 	for _, diskCache := range c.diskCaches {

weed/util/compression.go

@@ -126,7 +126,7 @@ func IsZstdContent(data []byte) bool {
 
 	// by file name extension
 	switch ext {
-	case ".zip", ".rar", ".gz", ".bz2", ".xz", ".zst":
+	case ".zip", ".rar", ".gz", ".bz2", ".xz", ".zst", ".br":
 		return false, true
 	case ".pdf", ".txt", ".html", ".htm", ".css", ".js", ".json":
 		return true, true
@@ -147,6 +147,9 @@ func IsZstdContent(data []byte) bool {
 		if strings.HasSuffix(mtype, "script") {
 			return true, true
 		}
+		if strings.HasSuffix(mtype, "vnd.rar") {
+			return false, true
+		}
 	}
 
 	if strings.HasPrefix(mtype, "audio/") {

weed/util/constants.go

@@ -5,7 +5,7 @@ import (
 )
 
 var (
-	VERSION = fmt.Sprintf("%s %d.%02d", sizeLimit, 2, 41)
+	VERSION = fmt.Sprintf("%s %d.%02d", sizeLimit, 2, 43)
 	COMMIT  = ""
 )
 

weed/util/minfreespace.go

@@ -0,0 +1,90 @@
+package util
+
+import (
+	"errors"
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"strconv"
+	"strings"
+)
+
+// MinFreeSpaceType is the type of MinFreeSpace.
+type MinFreeSpaceType int
+
+const (
+	// AsPercent set the MinFreeSpaceType to a percentage value from 0 to 100.
+	AsPercent MinFreeSpaceType = iota
+	// AsBytes set the MinFreeSpaceType to a absolute value bytes.
+	AsBytes
+)
+
+// MinFreeSpace is type that defines the limit for the minimum free space.
+type MinFreeSpace struct {
+	Type    MinFreeSpaceType
+	Bytes   uint64
+	Percent float32
+	Raw     string
+}
+
+// IsLow tells whether the free space is low or not.
+func (s MinFreeSpace) IsLow(freeBytes uint64, freePercent float32) (yes bool, desc string) {
+	switch s.Type {
+	case AsPercent:
+		yes = freePercent < s.Percent
+		op := IfElse(yes, "<", ">=")
+		return yes, fmt.Sprintf("disk free %.2f%% %s required %.2f%%", freePercent, op, s.Percent)
+	case AsBytes:
+		yes = freeBytes < s.Bytes
+		op := IfElse(yes, "<", ">=")
+		return yes, fmt.Sprintf("disk free %s %s required %s",
+			BytesToHumanReadable(freeBytes), op, BytesToHumanReadable(s.Bytes))
+	}
+
+	return false, ""
+}
+
+// String returns a string representation of MinFreeSpace.
+func (s MinFreeSpace) String() string {
+	switch s.Type {
+	case AsPercent:
+		return fmt.Sprintf("%.2f%%", s.Percent)
+	default:
+		return s.Raw
+	}
+}
+
+// MustParseMinFreeSpace parses comma-separated argument for min free space setting.
+// minFreeSpace has the high priority than minFreeSpacePercent if it is set.
+func MustParseMinFreeSpace(minFreeSpace string, minFreeSpacePercent string) (spaces []MinFreeSpace) {
+	ss := strings.Split(EmptyTo(minFreeSpace, minFreeSpacePercent), ",")
+	for _, freeString := range ss {
+		if vv, e := ParseMinFreeSpace(freeString); e == nil {
+			spaces = append(spaces, *vv)
+		} else {
+			glog.Fatalf("The value specified in -minFreeSpace not a valid value %s", freeString)
+		}
+	}
+
+	return spaces
+}
+
+var ErrMinFreeSpaceBadValue = errors.New("minFreeSpace is invalid")
+
+// ParseMinFreeSpace parses min free space expression s as percentage like 1,10 or human readable size like 10G
+func ParseMinFreeSpace(s string) (*MinFreeSpace, error) {
+	if percent, e := strconv.ParseFloat(s, 32); e == nil {
+		if percent < 0 || percent > 100 {
+			return nil, ErrMinFreeSpaceBadValue
+		}
+		return &MinFreeSpace{Type: AsPercent, Percent: float32(percent), Raw: s}, nil
+	}
+
+	if directSize, e := ParseBytes(s); e == nil {
+		if directSize <= 100 {
+			return nil, ErrMinFreeSpaceBadValue
+		}
+		return &MinFreeSpace{Type: AsBytes, Bytes: directSize, Raw: s}, nil
+	}
+
+	return nil, ErrMinFreeSpaceBadValue
+}

weed/util/minfreespace_test.go

@@ -0,0 +1,29 @@
+package util
+
+import "testing"
+
+func TestParseMinFreeSpace(t *testing.T) {
+	tests := []struct {
+		in    string
+		ok    bool
+		value *MinFreeSpace
+	}{
+		{in: "42", ok: true, value: &MinFreeSpace{Type: AsPercent, Percent: 42, Raw: "42"}},
+		{in: "-1", ok: false, value: nil},
+		{in: "101", ok: false, value: nil},
+		{in: "100B", ok: false, value: nil},
+		{in: "100Ki", ok: true, value: &MinFreeSpace{Type: AsBytes, Bytes: 100 * 1024, Raw: "100Ki"}},
+		{in: "100GiB", ok: true, value: &MinFreeSpace{Type: AsBytes, Bytes: 100 * 1024 * 1024 * 1024, Raw: "100GiB"}},
+		{in: "42M", ok: true, value: &MinFreeSpace{Type: AsBytes, Bytes: 42 * 1000 * 1000, Raw: "42M"}},
+	}
+
+	for _, p := range tests {
+		got, err := ParseMinFreeSpace(p.in)
+		if p.ok != (err == nil) {
+			t.Errorf("failed to test %v", p.in)
+		}
+		if p.ok && err == nil && *got != *p.value {
+			t.Errorf("failed to test %v", p.in)
+		}
+	}
+}