seaweedfs/weed/filer/reader_at.go

277 lines
7.9 KiB
Go
Raw Normal View History

2020-09-01 07:21:19 +00:00
package filer
import (
"context"
"fmt"
2020-11-01 10:36:43 +00:00
"io"
"math/rand"
"sync"
"github.com/chrislusf/seaweedfs/weed/glog"
"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
2020-11-01 08:58:48 +00:00
"github.com/chrislusf/seaweedfs/weed/util"
2020-04-11 19:45:24 +00:00
"github.com/chrislusf/seaweedfs/weed/util/chunk_cache"
"github.com/chrislusf/seaweedfs/weed/wdclient"
2020-10-08 06:58:32 +00:00
"github.com/golang/groupcache/singleflight"
2020-10-11 03:09:43 +00:00
)
type ChunkReadAt struct {
masterClient *wdclient.MasterClient
chunkViews []*ChunkView
lookupFileId wdclient.LookupFileIdFunctionType
readerLock sync.Mutex
2020-08-16 07:49:08 +00:00
fileSize int64
fetchGroup singleflight.Group
chunkCache chunk_cache.ChunkCache
lastChunkFileId string
lastChunkData []byte
readerPattern *ReaderPattern
}
var _ = io.ReaderAt(&ChunkReadAt{})
var _ = io.Closer(&ChunkReadAt{})
func LookupFn(filerClient filer_pb.FilerClient) wdclient.LookupFileIdFunctionType {
2020-10-04 03:16:42 +00:00
vidCache := make(map[string]*filer_pb.Locations)
2020-10-22 02:28:59 +00:00
var vicCacheLock sync.RWMutex
return func(fileId string) (targetUrls []string, err error) {
2020-10-04 03:16:42 +00:00
vid := VolumeId(fileId)
2020-10-22 02:28:59 +00:00
vicCacheLock.RLock()
2020-10-04 03:16:42 +00:00
locations, found := vidCache[vid]
2020-10-22 02:28:59 +00:00
vicCacheLock.RUnlock()
2020-10-04 03:16:42 +00:00
2020-11-01 08:58:48 +00:00
if !found {
2020-11-01 10:36:43 +00:00
util.Retry("lookup volume "+vid, func() error {
err = filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
2020-11-01 08:58:48 +00:00
resp, err := client.LookupVolume(context.Background(), &filer_pb.LookupVolumeRequest{
VolumeIds: []string{vid},
})
if err != nil {
return err
}
locations = resp.LocationsMap[vid]
if locations == nil || len(locations.Locations) == 0 {
glog.V(0).Infof("failed to locate %s", fileId)
return fmt.Errorf("failed to locate %s", fileId)
}
vicCacheLock.Lock()
vidCache[vid] = locations
vicCacheLock.Unlock()
return nil
2020-10-04 03:16:42 +00:00
})
2020-11-01 08:58:48 +00:00
return err
})
2020-10-04 03:16:42 +00:00
}
2020-10-10 22:43:22 +00:00
if err != nil {
return nil, err
}
for _, loc := range locations.Locations {
2021-01-28 22:36:29 +00:00
volumeServerAddress := filerClient.AdjustedUrl(loc)
targetUrl := fmt.Sprintf("http://%s/%s", volumeServerAddress, fileId)
targetUrls = append(targetUrls, targetUrl)
}
2020-10-08 06:58:32 +00:00
for i := len(targetUrls) - 1; i > 0; i-- {
j := rand.Intn(i + 1)
targetUrls[i], targetUrls[j] = targetUrls[j], targetUrls[i]
}
return
}
}
func NewChunkReaderAtFromClient(lookupFn wdclient.LookupFileIdFunctionType, chunkViews []*ChunkView, chunkCache chunk_cache.ChunkCache, fileSize int64) *ChunkReadAt {
return &ChunkReadAt{
chunkViews: chunkViews,
lookupFileId: lookupFn,
chunkCache: chunkCache,
fileSize: fileSize,
readerPattern: NewReaderPattern(),
}
}
func (c *ChunkReadAt) Close() error {
c.lastChunkData = nil
c.lastChunkFileId = ""
return nil
}
func (c *ChunkReadAt) ReadAt(p []byte, offset int64) (n int, err error) {
c.readerPattern.MonitorReadAt(offset, len(p))
c.readerLock.Lock()
defer c.readerLock.Unlock()
2021-05-11 04:47:51 +00:00
// glog.V(4).Infof("ReadAt [%d,%d) of total file size %d bytes %d chunk views", offset, offset+int64(len(p)), c.fileSize, len(c.chunkViews))
2021-01-18 09:14:27 +00:00
return c.doReadAt(p, offset)
}
func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, err error) {
2020-08-18 03:14:40 +00:00
startOffset, remaining := offset, int64(len(p))
2020-10-04 08:31:04 +00:00
var nextChunk *ChunkView
2020-08-17 23:05:40 +00:00
for i, chunk := range c.chunkViews {
2020-08-18 03:14:40 +00:00
if remaining <= 0 {
break
}
2020-10-04 08:31:04 +00:00
if i+1 < len(c.chunkViews) {
nextChunk = c.chunkViews[i+1]
} else {
nextChunk = nil
}
2020-08-18 03:14:40 +00:00
if startOffset < chunk.LogicOffset {
gap := int(chunk.LogicOffset - startOffset)
2020-08-18 07:32:01 +00:00
glog.V(4).Infof("zero [%d,%d)", startOffset, startOffset+int64(gap))
n += int(min(int64(gap), remaining))
2020-08-18 03:14:40 +00:00
startOffset, remaining = chunk.LogicOffset, remaining-int64(gap)
if remaining <= 0 {
break
}
2020-08-17 04:07:46 +00:00
}
2020-08-16 22:16:46 +00:00
// fmt.Printf(">>> doReadAt [%d,%d), chunk[%d,%d)\n", offset, offset+int64(len(p)), chunk.LogicOffset, chunk.LogicOffset+int64(chunk.Size))
2020-08-18 03:14:40 +00:00
chunkStart, chunkStop := max(chunk.LogicOffset, startOffset), min(chunk.LogicOffset+int64(chunk.Size), startOffset+remaining)
2020-08-17 04:07:46 +00:00
if chunkStart >= chunkStop {
continue
}
2021-05-11 04:47:51 +00:00
// 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
bufferOffset := chunkStart - chunk.LogicOffset + chunk.Offset
bufferLength := chunkStop - chunkStart
buffer, err = c.readChunkSlice(chunk, nextChunk, uint64(bufferOffset), uint64(bufferLength))
2020-08-17 04:07:46 +00:00
if err != nil {
glog.Errorf("fetching chunk %+v: %v\n", chunk, err)
return
}
copied := copy(p[startOffset-offset:chunkStop-chunkStart+startOffset-offset], buffer)
2020-08-17 04:07:46 +00:00
n += copied
2020-08-18 03:20:08 +00:00
startOffset, remaining = startOffset+int64(copied), remaining-int64(copied)
}
2020-08-16 07:49:08 +00:00
2021-05-11 04:47:51 +00:00
// glog.V(4).Infof("doReadAt [%d,%d), n:%v, err:%v", offset, offset+int64(len(p)), n, err)
2020-08-16 07:49:08 +00:00
2020-08-18 15:50:14 +00:00
if err == nil && remaining > 0 && c.fileSize > startOffset {
2020-08-30 05:28:33 +00:00
delta := int(min(remaining, c.fileSize-startOffset))
2020-08-18 07:34:15 +00:00
glog.V(4).Infof("zero2 [%d,%d) of file size %d bytes", startOffset, startOffset+int64(delta), c.fileSize)
2020-08-18 05:46:32 +00:00
n += delta
}
2020-08-18 03:14:40 +00:00
if err == nil && offset+int64(len(p)) >= c.fileSize {
2020-08-16 07:49:08 +00:00
err = io.EOF
}
2020-08-16 22:16:46 +00:00
// fmt.Printf("~~~ filled %d, err: %v\n\n", n, err)
return
}
func (c *ChunkReadAt) readChunkSlice(chunkView *ChunkView, nextChunkViews *ChunkView, offset, length uint64) ([]byte, error) {
var chunkSlice []byte
if chunkView.LogicOffset == 0 {
chunkSlice = c.chunkCache.GetChunkSlice(chunkView.FileId, offset, length)
}
if len(chunkSlice) > 0 {
return chunkSlice, nil
}
2021-12-23 00:05:08 +00:00
if c.lookupFileId == nil {
return nil, nil
}
2021-12-23 00:17:30 +00:00
if c.readerPattern.IsRandomMode() {
return c.doFetchRangeChunkData(chunkView, offset, length)
}
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) {
2020-10-04 03:16:42 +00:00
if c.lastChunkFileId == chunkView.FileId {
return c.lastChunkData, nil
}
2020-10-04 08:31:04 +00:00
v, doErr := c.readOneWholeChunk(chunkView)
2020-10-04 08:31:04 +00:00
if doErr != nil {
2020-10-09 06:19:20 +00:00
return nil, doErr
2020-04-12 08:13:57 +00:00
}
2020-10-04 08:31:04 +00:00
chunkData = v.([]byte)
c.lastChunkData = chunkData
c.lastChunkFileId = chunkView.FileId
2020-10-05 21:06:18 +00:00
for _, nextChunkView := range nextChunkViews {
if c.chunkCache != nil && nextChunkView != nil {
go c.readOneWholeChunk(nextChunkView)
2020-10-04 08:31:04 +00:00
}
2020-10-05 21:06:18 +00:00
}
2020-10-04 08:31:04 +00:00
2020-08-16 22:16:46 +00:00
return
}
2020-10-04 08:31:04 +00:00
func (c *ChunkReadAt) readOneWholeChunk(chunkView *ChunkView) (interface{}, error) {
var err error
return c.fetchGroup.Do(chunkView.FileId, func() (interface{}, error) {
glog.V(4).Infof("readFromWholeChunkData %s offset %d [%d,%d) size at least %d", chunkView.FileId, chunkView.Offset, chunkView.LogicOffset, chunkView.LogicOffset+int64(chunkView.Size), chunkView.ChunkSize)
var data []byte
if chunkView.LogicOffset == 0 {
data = c.chunkCache.GetChunk(chunkView.FileId, chunkView.ChunkSize)
}
2020-10-04 08:31:04 +00:00
if data != nil {
glog.V(4).Infof("cache hit %s [%d,%d)", chunkView.FileId, chunkView.LogicOffset-chunkView.Offset, chunkView.LogicOffset-chunkView.Offset+int64(len(data)))
} else {
var err error
data, err = c.doFetchFullChunkData(chunkView)
if err != nil {
return data, err
}
if chunkView.LogicOffset == 0 {
// only cache the first chunk
c.chunkCache.SetChunk(chunkView.FileId, data)
}
2020-10-04 08:31:04 +00:00
}
return data, err
})
}
2020-10-04 03:16:42 +00:00
func (c *ChunkReadAt) doFetchFullChunkData(chunkView *ChunkView) ([]byte, error) {
2020-10-04 08:31:04 +00:00
2020-10-04 23:21:43 +00:00
glog.V(4).Infof("+ doFetchFullChunkData %s", chunkView.FileId)
2020-10-04 08:31:04 +00:00
2020-10-04 03:16:42 +00:00
data, err := fetchChunk(c.lookupFileId, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped)
2020-10-04 23:21:43 +00:00
glog.V(4).Infof("- doFetchFullChunkData %s", chunkView.FileId)
2020-10-04 08:31:04 +00:00
2020-10-04 03:16:42 +00:00
return data, err
}
func (c *ChunkReadAt) doFetchRangeChunkData(chunkView *ChunkView, offset, length uint64) ([]byte, error) {
glog.V(4).Infof("+ doFetchFullChunkData %s", chunkView.FileId)
data, err := fetchChunkRange(c.lookupFileId, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped, int64(offset), int(length))
glog.V(4).Infof("- doFetchFullChunkData %s", chunkView.FileId)
return data, err
}