seaweedfs/weed/server/filer_server_handlers_write_autochunk.go

406 lines
12 KiB
Go
Raw Normal View History

2018-05-28 06:53:10 +00:00
package weed_server
import (
2019-03-15 22:55:34 +00:00
"context"
"crypto/md5"
"fmt"
2020-08-08 17:45:37 +00:00
"hash"
2018-05-28 06:53:10 +00:00
"io"
"io/ioutil"
2018-05-28 06:53:10 +00:00
"net/http"
"os"
2018-05-28 06:53:10 +00:00
"path"
"strconv"
"strings"
2018-05-28 06:59:49 +00:00
"time"
2020-09-01 07:21:19 +00:00
"github.com/chrislusf/seaweedfs/weed/filer"
2018-05-28 06:53:10 +00:00
"github.com/chrislusf/seaweedfs/weed/glog"
"github.com/chrislusf/seaweedfs/weed/operation"
"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
2020-10-29 08:05:40 +00:00
xhttp "github.com/chrislusf/seaweedfs/weed/s3api/http"
2019-02-15 08:09:19 +00:00
"github.com/chrislusf/seaweedfs/weed/security"
2019-06-23 05:53:52 +00:00
"github.com/chrislusf/seaweedfs/weed/stats"
"github.com/chrislusf/seaweedfs/weed/storage/needle"
2020-03-23 07:01:34 +00:00
"github.com/chrislusf/seaweedfs/weed/util"
2018-05-28 06:53:10 +00:00
)
2020-11-16 00:58:48 +00:00
func (fs *FilerServer) autoChunk(ctx context.Context, w http.ResponseWriter, r *http.Request, so *operation.StorageOption) {
2018-05-28 06:53:10 +00:00
// autoChunking can be set at the command-line level or as a query param. Query param overrides command-line
query := r.URL.Query()
parsedMaxMB, _ := strconv.ParseInt(query.Get("maxMB"), 10, 32)
maxMB := int32(parsedMaxMB)
2018-07-07 09:18:47 +00:00
if maxMB <= 0 && fs.option.MaxMB > 0 {
maxMB = int32(fs.option.MaxMB)
2018-05-28 06:53:10 +00:00
}
chunkSize := 1024 * 1024 * maxMB
2020-08-08 17:45:37 +00:00
stats.FilerRequestCounter.WithLabelValues("postAutoChunk").Inc()
start := time.Now()
defer func() {
stats.FilerRequestHistogram.WithLabelValues("postAutoChunk").Observe(time.Since(start).Seconds())
}()
var reply *FilerPostResult
var err error
var md5bytes []byte
2020-08-08 17:45:37 +00:00
if r.Method == "POST" {
if r.Header.Get("Content-Type") == "" && strings.HasSuffix(r.URL.Path, "/") {
reply, err = fs.mkdir(ctx, w, r)
} else {
2020-11-15 22:41:56 +00:00
reply, md5bytes, err = fs.doPostAutoChunk(ctx, w, r, chunkSize, so)
}
2020-08-08 17:45:37 +00:00
} else {
2020-11-15 22:41:56 +00:00
reply, md5bytes, err = fs.doPutAutoChunk(ctx, w, r, chunkSize, so)
2020-08-08 17:45:37 +00:00
}
2018-05-28 06:53:10 +00:00
if err != nil {
if strings.HasPrefix(err.Error(), "read input:") {
writeJsonError(w, r, 499, err)
} else {
writeJsonError(w, r, http.StatusInternalServerError, err)
}
2018-05-28 06:53:10 +00:00
} else if reply != nil {
if len(md5bytes) > 0 {
w.Header().Set("Content-MD5", util.Base64Encode(md5bytes))
}
2018-05-28 06:53:10 +00:00
writeJsonQuiet(w, r, http.StatusCreated, reply)
}
}
2020-11-16 00:58:48 +00:00
func (fs *FilerServer) doPostAutoChunk(ctx context.Context, w http.ResponseWriter, r *http.Request, chunkSize int32, so *operation.StorageOption) (filerResult *FilerPostResult, md5bytes []byte, replyerr error) {
2018-05-28 06:53:10 +00:00
multipartReader, multipartReaderErr := r.MultipartReader()
if multipartReaderErr != nil {
return nil, nil, multipartReaderErr
2018-05-28 06:53:10 +00:00
}
part1, part1Err := multipartReader.NextPart()
if part1Err != nil {
return nil, nil, part1Err
2018-05-28 06:53:10 +00:00
}
fileName := part1.FileName()
if fileName != "" {
fileName = path.Base(fileName)
}
contentType := part1.Header.Get("Content-Type")
2020-08-08 17:45:37 +00:00
if contentType == "application/octet-stream" {
contentType = ""
}
2020-11-30 12:34:04 +00:00
fileChunks, md5Hash, chunkOffset, err, smallContent := fs.uploadReaderToChunks(w, r, part1, chunkSize, fileName, contentType, so)
2020-08-08 17:45:37 +00:00
if err != nil {
return nil, nil, err
2020-08-08 17:45:37 +00:00
}
2018-05-28 06:53:10 +00:00
md5bytes = md5Hash.Sum(nil)
2020-11-30 12:34:04 +00:00
filerResult, replyerr = fs.saveMetaData(ctx, r, fileName, contentType, so, md5bytes, fileChunks, chunkOffset, smallContent)
2018-05-28 06:53:10 +00:00
2020-08-08 17:45:37 +00:00
return
}
2018-05-28 06:53:10 +00:00
2020-11-16 00:58:48 +00:00
func (fs *FilerServer) doPutAutoChunk(ctx context.Context, w http.ResponseWriter, r *http.Request, chunkSize int32, so *operation.StorageOption) (filerResult *FilerPostResult, md5bytes []byte, replyerr error) {
fileName := path.Base(r.URL.Path)
contentType := r.Header.Get("Content-Type")
if contentType == "application/octet-stream" {
contentType = ""
}
2020-11-30 12:34:04 +00:00
fileChunks, md5Hash, chunkOffset, err, smallContent := fs.uploadReaderToChunks(w, r, r.Body, chunkSize, fileName, contentType, so)
2020-08-08 17:45:37 +00:00
if err != nil {
return nil, nil, err
2018-05-28 06:53:10 +00:00
}
md5bytes = md5Hash.Sum(nil)
2020-11-30 12:34:04 +00:00
filerResult, replyerr = fs.saveMetaData(ctx, r, fileName, contentType, so, md5bytes, fileChunks, chunkOffset, smallContent)
2020-08-08 17:45:37 +00:00
return
}
2021-01-20 21:40:32 +00:00
func isAppend(r *http.Request) bool {
return r.URL.Query().Get("op") == "append"
}
2020-11-30 12:34:04 +00:00
func (fs *FilerServer) saveMetaData(ctx context.Context, r *http.Request, fileName string, contentType string, so *operation.StorageOption, md5bytes []byte, fileChunks []*filer_pb.FileChunk, chunkOffset int64, content []byte) (filerResult *FilerPostResult, replyerr error) {
// detect file mode
modeStr := r.URL.Query().Get("mode")
if modeStr == "" {
modeStr = "0660"
}
mode, err := strconv.ParseUint(modeStr, 8, 32)
if err != nil {
glog.Errorf("Invalid mode format: %s, use 0660 by default", modeStr)
mode = 0660
}
// fix the path
2018-05-28 06:53:10 +00:00
path := r.URL.Path
if strings.HasSuffix(path, "/") {
if fileName != "" {
path += fileName
2018-05-28 06:53:10 +00:00
}
}
2021-01-20 21:40:32 +00:00
var entry *filer.Entry
var mergedChunks []*filer_pb.FileChunk
// when it is an append
if isAppend(r) {
existingEntry, findErr := fs.filer.FindEntry(ctx, util.FullPath(path))
if findErr != nil && findErr != filer_pb.ErrNotFound {
glog.V(0).Infof("failing to find %s: %v", path, findErr)
}
entry = existingEntry
2018-05-28 06:53:10 +00:00
}
2021-01-20 21:40:32 +00:00
if entry != nil {
entry.Mtime = time.Now()
entry.Md5 = nil
// adjust chunk offsets
for _, chunk := range fileChunks {
chunk.Offset += int64(entry.FileSize)
}
mergedChunks = append(entry.Chunks, fileChunks...)
entry.FileSize += uint64(chunkOffset)
// TODO
if len(entry.Content) > 0 {
replyerr = fmt.Errorf("append to small file is not supported yet")
return
}
} else {
glog.V(4).Infoln("saving", path)
mergedChunks = fileChunks
entry = &filer.Entry{
FullPath: util.FullPath(path),
Attr: filer.Attr{
Mtime: time.Now(),
Crtime: time.Now(),
Mode: os.FileMode(mode),
Uid: OS_UID,
Gid: OS_GID,
Replication: so.Replication,
Collection: so.Collection,
TtlSec: so.TtlSeconds,
DiskType: so.DiskType,
2021-01-20 21:40:32 +00:00
Mime: contentType,
Md5: md5bytes,
FileSize: uint64(chunkOffset),
},
Content: content,
}
}
// maybe compact entry chunks
mergedChunks, replyerr = filer.MaybeManifestize(fs.saveAsChunk(so), mergedChunks)
if replyerr != nil {
glog.V(0).Infof("manifestize %s: %v", r.RequestURI, replyerr)
return
2018-05-28 06:53:10 +00:00
}
2021-01-20 21:40:32 +00:00
entry.Chunks = mergedChunks
filerResult = &FilerPostResult{
Name: fileName,
Size: int64(entry.FileSize),
}
if entry.Extended == nil {
entry.Extended = make(map[string][]byte)
}
2020-10-28 10:16:05 +00:00
fs.saveAmzMetaData(r, entry)
for k, v := range r.Header {
if len(v) > 0 && strings.HasPrefix(k, needle.PairNamePrefix) {
2020-11-03 08:21:10 +00:00
entry.Extended[k] = []byte(v[0])
}
}
if dbErr := fs.filer.CreateEntry(ctx, entry, false, false, nil); dbErr != nil {
2021-01-20 21:40:32 +00:00
fs.filer.DeleteChunks(fileChunks)
2019-06-23 05:53:52 +00:00
replyerr = dbErr
filerResult.Error = dbErr.Error()
glog.V(0).Infof("failing to write %s to filer server : %v", path, dbErr)
2018-05-28 06:53:10 +00:00
}
2020-08-08 17:45:37 +00:00
return filerResult, replyerr
}
2018-05-28 06:53:10 +00:00
2020-11-30 12:34:04 +00:00
func (fs *FilerServer) uploadReaderToChunks(w http.ResponseWriter, r *http.Request, reader io.Reader, chunkSize int32, fileName, contentType string, so *operation.StorageOption) ([]*filer_pb.FileChunk, hash.Hash, int64, error, []byte) {
2020-08-08 17:45:37 +00:00
var fileChunks []*filer_pb.FileChunk
md5Hash := md5.New()
var partReader = ioutil.NopCloser(io.TeeReader(reader, md5Hash))
chunkOffset := int64(0)
var smallContent []byte
2020-08-08 17:45:37 +00:00
for {
2020-08-08 17:45:37 +00:00
limitedReader := io.LimitReader(partReader, int64(chunkSize))
data, err := ioutil.ReadAll(limitedReader)
if err != nil {
return nil, nil, 0, err, nil
}
2021-01-20 21:40:32 +00:00
if chunkOffset == 0 && !isAppend(r) {
if len(data) < fs.option.SaveToFilerLimit || strings.HasPrefix(r.URL.Path, filer.DirectoryEtcRoot) && len(data) < 4*1024 {
smallContent = data
chunkOffset += int64(len(data))
break
}
}
dataReader := util.NewBytesReader(data)
// retry to assign a different file id
var fileId, urlLocation string
var auth security.EncodedJwt
var assignErr, uploadErr error
var uploadResult *operation.UploadResult
for i := 0; i < 3; i++ {
// assign one file id for one chunk
fileId, urlLocation, auth, assignErr = fs.assignNewFileInfo(so)
if assignErr != nil {
return nil, nil, 0, assignErr, nil
}
// upload the chunk to the volume server
uploadResult, uploadErr, _ = fs.doUpload(urlLocation, w, r, dataReader, fileName, contentType, nil, auth)
if uploadErr != nil {
time.Sleep(251 * time.Millisecond)
continue
}
break
2020-08-08 17:45:37 +00:00
}
if uploadErr != nil {
2020-11-30 12:34:04 +00:00
return nil, nil, 0, uploadErr, nil
2020-08-08 17:45:37 +00:00
}
// if last chunk exhausted the reader exactly at the border
if uploadResult.Size == 0 {
break
}
// Save to chunk manifest structure
fileChunks = append(fileChunks, uploadResult.ToPbFileChunk(fileId, chunkOffset))
glog.V(4).Infof("uploaded %s chunk %d to %s [%d,%d)", fileName, len(fileChunks), fileId, chunkOffset, chunkOffset+int64(uploadResult.Size))
2020-08-08 17:45:37 +00:00
// reset variables for the next chunk
chunkOffset = chunkOffset + int64(uploadResult.Size)
// if last chunk was not at full chunk size, but already exhausted the reader
if int64(uploadResult.Size) < int64(chunkSize) {
break
}
}
2020-11-30 12:34:04 +00:00
return fileChunks, md5Hash, chunkOffset, nil, smallContent
2018-05-28 06:53:10 +00:00
}
2020-11-30 12:34:04 +00:00
func (fs *FilerServer) doUpload(urlLocation string, w http.ResponseWriter, r *http.Request, limitedReader io.Reader, fileName string, contentType string, pairMap map[string]string, auth security.EncodedJwt) (*operation.UploadResult, error, []byte) {
2018-05-28 06:53:10 +00:00
2019-06-23 05:53:52 +00:00
stats.FilerRequestCounter.WithLabelValues("postAutoChunkUpload").Inc()
start := time.Now()
2019-06-23 08:57:51 +00:00
defer func() {
stats.FilerRequestHistogram.WithLabelValues("postAutoChunkUpload").Observe(time.Since(start).Seconds())
}()
2019-06-23 05:53:52 +00:00
2020-11-30 12:34:04 +00:00
uploadResult, err, data := operation.Upload(urlLocation, fileName, fs.option.Cipher, limitedReader, false, contentType, pairMap, auth)
return uploadResult, err, data
2018-05-28 06:53:10 +00:00
}
2020-11-16 00:58:48 +00:00
func (fs *FilerServer) saveAsChunk(so *operation.StorageOption) filer.SaveDataAsChunkFunctionType {
return func(reader io.Reader, name string, offset int64) (*filer_pb.FileChunk, string, string, error) {
// assign one file id for one chunk
2020-11-15 22:41:56 +00:00
fileId, urlLocation, auth, assignErr := fs.assignNewFileInfo(so)
if assignErr != nil {
return nil, "", "", assignErr
}
// upload the chunk to the volume server
uploadResult, uploadErr, _ := operation.Upload(urlLocation, name, fs.option.Cipher, reader, false, "", nil, auth)
if uploadErr != nil {
return nil, "", "", uploadErr
}
2020-11-15 22:41:56 +00:00
return uploadResult.ToPbFileChunk(fileId, offset), so.Collection, so.Replication, nil
}
}
func (fs *FilerServer) mkdir(ctx context.Context, w http.ResponseWriter, r *http.Request) (filerResult *FilerPostResult, replyerr error) {
// detect file mode
modeStr := r.URL.Query().Get("mode")
if modeStr == "" {
modeStr = "0660"
}
mode, err := strconv.ParseUint(modeStr, 8, 32)
if err != nil {
glog.Errorf("Invalid mode format: %s, use 0660 by default", modeStr)
mode = 0660
}
// fix the path
path := r.URL.Path
if strings.HasSuffix(path, "/") {
path = path[:len(path)-1]
}
existingEntry, err := fs.filer.FindEntry(ctx, util.FullPath(path))
if err == nil && existingEntry != nil {
replyerr = fmt.Errorf("dir %s already exists", path)
return
}
glog.V(4).Infoln("mkdir", path)
entry := &filer.Entry{
FullPath: util.FullPath(path),
Attr: filer.Attr{
2020-09-16 08:27:24 +00:00
Mtime: time.Now(),
Crtime: time.Now(),
Mode: os.FileMode(mode) | os.ModeDir,
Uid: OS_UID,
Gid: OS_GID,
},
}
filerResult = &FilerPostResult{
Name: util.FullPath(path).Name(),
}
if dbErr := fs.filer.CreateEntry(ctx, entry, false, false, nil); dbErr != nil {
replyerr = dbErr
filerResult.Error = dbErr.Error()
glog.V(0).Infof("failing to create dir %s on filer server : %v", path, dbErr)
}
return filerResult, replyerr
}
2020-10-28 10:16:05 +00:00
func (fs *FilerServer) saveAmzMetaData(r *http.Request, entry *filer.Entry) {
2020-10-29 08:05:40 +00:00
if sc := r.Header.Get(xhttp.AmzStorageClass); sc != "" {
entry.Extended[xhttp.AmzStorageClass] = []byte(sc)
2020-10-28 10:16:05 +00:00
}
2020-10-29 08:05:40 +00:00
if tags := r.Header.Get(xhttp.AmzObjectTagging); tags != "" {
2020-10-28 10:16:05 +00:00
for _, v := range strings.Split(tags, "&") {
tag := strings.Split(v, "=")
if len(tag) == 2 {
2020-10-29 08:05:40 +00:00
entry.Extended[xhttp.AmzObjectTagging+"-"+tag[0]] = []byte(tag[1])
2020-10-28 10:16:05 +00:00
}
}
}
for header, values := range r.Header {
2020-10-29 08:05:40 +00:00
if strings.HasPrefix(header, xhttp.AmzUserMetaPrefix) {
2020-10-28 10:16:05 +00:00
for _, value := range values {
entry.Extended[header] = []byte(value)
}
}
}
}