2020-09-01 15:21:19 +08:00
|
|
|
package filer
|
2020-03-27 19:50:51 +08:00
|
|
|
|
|
|
|
import (
|
|
|
|
"context"
|
|
|
|
"fmt"
|
2020-11-01 18:36:43 +08:00
|
|
|
"io"
|
|
|
|
"math/rand"
|
|
|
|
"sync"
|
|
|
|
|
2020-03-27 19:50:51 +08:00
|
|
|
"github.com/chrislusf/seaweedfs/weed/glog"
|
|
|
|
"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
|
2020-11-01 16:58:48 +08:00
|
|
|
"github.com/chrislusf/seaweedfs/weed/util"
|
2020-04-12 03:45:24 +08:00
|
|
|
"github.com/chrislusf/seaweedfs/weed/util/chunk_cache"
|
2020-03-27 19:50:51 +08:00
|
|
|
"github.com/chrislusf/seaweedfs/weed/wdclient"
|
2020-10-11 11:09:43 +08:00
|
|
|
)
|
|
|
|
|
2020-03-27 19:50:51 +08:00
|
|
|
type ChunkReadAt struct {
|
2022-02-26 18:16:47 +08:00
|
|
|
masterClient *wdclient.MasterClient
|
|
|
|
chunkViews []*ChunkView
|
|
|
|
readerLock sync.Mutex
|
|
|
|
fileSize int64
|
|
|
|
readerCache *ReaderCache
|
|
|
|
readerPattern *ReaderPattern
|
2022-02-27 15:20:45 +08:00
|
|
|
lastChunkFid string
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
|
|
|
|
2020-12-09 14:26:46 +08:00
|
|
|
var _ = io.ReaderAt(&ChunkReadAt{})
|
|
|
|
var _ = io.Closer(&ChunkReadAt{})
|
2020-03-27 19:50:51 +08:00
|
|
|
|
2021-01-06 20:21:34 +08:00
|
|
|
func LookupFn(filerClient filer_pb.FilerClient) wdclient.LookupFileIdFunctionType {
|
2020-10-04 11:16:42 +08:00
|
|
|
|
|
|
|
vidCache := make(map[string]*filer_pb.Locations)
|
2020-10-22 10:28:59 +08:00
|
|
|
var vicCacheLock sync.RWMutex
|
2020-10-08 13:49:04 +08:00
|
|
|
return func(fileId string) (targetUrls []string, err error) {
|
2020-10-04 11:16:42 +08:00
|
|
|
vid := VolumeId(fileId)
|
2020-10-22 10:28:59 +08:00
|
|
|
vicCacheLock.RLock()
|
2020-10-04 11:16:42 +08:00
|
|
|
locations, found := vidCache[vid]
|
2020-10-22 10:28:59 +08:00
|
|
|
vicCacheLock.RUnlock()
|
2020-10-04 11:16:42 +08:00
|
|
|
|
2020-11-01 16:58:48 +08:00
|
|
|
if !found {
|
2020-11-01 18:36:43 +08:00
|
|
|
util.Retry("lookup volume "+vid, func() error {
|
2021-12-26 16:15:03 +08:00
|
|
|
err = filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
|
2020-11-01 16:58:48 +08:00
|
|
|
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 11:16:42 +08:00
|
|
|
})
|
2020-11-01 16:58:48 +08:00
|
|
|
return err
|
2020-04-30 08:40:08 +08:00
|
|
|
})
|
2020-10-04 11:16:42 +08:00
|
|
|
}
|
2020-04-30 08:40:08 +08:00
|
|
|
|
2020-10-11 06:43:22 +08:00
|
|
|
if err != nil {
|
|
|
|
return nil, err
|
|
|
|
}
|
|
|
|
|
2020-10-08 13:49:04 +08:00
|
|
|
for _, loc := range locations.Locations {
|
2021-01-29 06:36:29 +08:00
|
|
|
volumeServerAddress := filerClient.AdjustedUrl(loc)
|
2020-10-08 13:49:04 +08:00
|
|
|
targetUrl := fmt.Sprintf("http://%s/%s", volumeServerAddress, fileId)
|
|
|
|
targetUrls = append(targetUrls, targetUrl)
|
|
|
|
}
|
2020-04-30 08:40:08 +08:00
|
|
|
|
2020-10-08 14:58:32 +08:00
|
|
|
for i := len(targetUrls) - 1; i > 0; i-- {
|
|
|
|
j := rand.Intn(i + 1)
|
|
|
|
targetUrls[i], targetUrls[j] = targetUrls[j], targetUrls[i]
|
|
|
|
}
|
|
|
|
|
2020-04-30 08:40:08 +08:00
|
|
|
return
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-01-25 11:01:58 +08:00
|
|
|
func NewChunkReaderAtFromClient(lookupFn wdclient.LookupFileIdFunctionType, chunkViews []*ChunkView, chunkCache chunk_cache.ChunkCache, fileSize int64) *ChunkReadAt {
|
2020-03-27 19:50:51 +08:00
|
|
|
|
|
|
|
return &ChunkReadAt{
|
2021-12-20 14:43:14 +08:00
|
|
|
chunkViews: chunkViews,
|
|
|
|
fileSize: fileSize,
|
2022-02-27 15:20:45 +08:00
|
|
|
readerCache: newReaderCache(32, chunkCache, lookupFn),
|
2021-12-20 14:43:14 +08:00
|
|
|
readerPattern: NewReaderPattern(),
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-09 14:26:46 +08:00
|
|
|
func (c *ChunkReadAt) Close() error {
|
2022-02-26 18:16:47 +08:00
|
|
|
c.readerCache.destroy()
|
2020-12-09 14:26:46 +08:00
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
2020-03-27 19:50:51 +08:00
|
|
|
func (c *ChunkReadAt) ReadAt(p []byte, offset int64) (n int, err error) {
|
|
|
|
|
2021-12-29 08:30:33 +08:00
|
|
|
c.readerPattern.MonitorReadAt(offset, len(p))
|
|
|
|
|
2020-03-27 19:50:51 +08:00
|
|
|
c.readerLock.Lock()
|
|
|
|
defer c.readerLock.Unlock()
|
|
|
|
|
2021-05-11 12:47:51 +08:00
|
|
|
// 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 17:14:27 +08:00
|
|
|
return c.doReadAt(p, offset)
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, err error) {
|
|
|
|
|
2020-08-18 11:14:40 +08:00
|
|
|
startOffset, remaining := offset, int64(len(p))
|
2022-02-26 19:06:17 +08:00
|
|
|
var nextChunks []*ChunkView
|
2020-08-18 07:05:40 +08:00
|
|
|
for i, chunk := range c.chunkViews {
|
2020-08-18 11:14:40 +08:00
|
|
|
if remaining <= 0 {
|
|
|
|
break
|
|
|
|
}
|
2020-10-04 16:31:04 +08:00
|
|
|
if i+1 < len(c.chunkViews) {
|
2022-02-26 19:06:17 +08:00
|
|
|
nextChunks = c.chunkViews[i+1:]
|
2020-10-04 16:31:04 +08:00
|
|
|
}
|
2020-08-18 11:14:40 +08:00
|
|
|
if startOffset < chunk.LogicOffset {
|
|
|
|
gap := int(chunk.LogicOffset - startOffset)
|
2022-01-13 22:20:33 +08:00
|
|
|
glog.V(4).Infof("zero [%d,%d)", startOffset, chunk.LogicOffset)
|
2020-08-18 15:32:01 +08:00
|
|
|
n += int(min(int64(gap), remaining))
|
2020-08-18 11:14:40 +08:00
|
|
|
startOffset, remaining = chunk.LogicOffset, remaining-int64(gap)
|
|
|
|
if remaining <= 0 {
|
|
|
|
break
|
|
|
|
}
|
2020-08-17 12:07:46 +08:00
|
|
|
}
|
2020-08-17 06:16:46 +08:00
|
|
|
// fmt.Printf(">>> doReadAt [%d,%d), chunk[%d,%d)\n", offset, offset+int64(len(p)), chunk.LogicOffset, chunk.LogicOffset+int64(chunk.Size))
|
2020-08-18 11:14:40 +08:00
|
|
|
chunkStart, chunkStop := max(chunk.LogicOffset, startOffset), min(chunk.LogicOffset+int64(chunk.Size), startOffset+remaining)
|
2020-08-17 12:07:46 +08:00
|
|
|
if chunkStart >= chunkStop {
|
|
|
|
continue
|
|
|
|
}
|
2021-05-11 12:47:51 +08:00
|
|
|
// 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))
|
2021-04-29 07:13:37 +08:00
|
|
|
bufferOffset := chunkStart - chunk.LogicOffset + chunk.Offset
|
2022-02-26 19:06:17 +08:00
|
|
|
copied, err := c.readChunkSliceAt(p[startOffset-offset:chunkStop-chunkStart+startOffset-offset], chunk, nextChunks, uint64(bufferOffset))
|
2020-08-17 12:07:46 +08:00
|
|
|
if err != nil {
|
|
|
|
glog.Errorf("fetching chunk %+v: %v\n", chunk, err)
|
2022-02-26 18:16:47 +08:00
|
|
|
return copied, err
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
2021-04-29 07:13:37 +08:00
|
|
|
|
2020-08-17 12:07:46 +08:00
|
|
|
n += copied
|
2020-08-18 11:20:08 +08:00
|
|
|
startOffset, remaining = startOffset+int64(copied), remaining-int64(copied)
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
2020-08-16 15:49:08 +08:00
|
|
|
|
2021-05-11 12:47:51 +08:00
|
|
|
// glog.V(4).Infof("doReadAt [%d,%d), n:%v, err:%v", offset, offset+int64(len(p)), n, err)
|
2020-08-16 15:49:08 +08:00
|
|
|
|
2020-08-18 23:50:14 +08:00
|
|
|
if err == nil && remaining > 0 && c.fileSize > startOffset {
|
2020-08-30 13:28:33 +08:00
|
|
|
delta := int(min(remaining, c.fileSize-startOffset))
|
2020-08-18 15:34:15 +08:00
|
|
|
glog.V(4).Infof("zero2 [%d,%d) of file size %d bytes", startOffset, startOffset+int64(delta), c.fileSize)
|
2020-08-18 13:46:32 +08:00
|
|
|
n += delta
|
2020-03-27 19:50:51 +08:00
|
|
|
}
|
2020-08-18 11:14:40 +08:00
|
|
|
|
2020-10-15 03:18:24 +08:00
|
|
|
if err == nil && offset+int64(len(p)) >= c.fileSize {
|
2020-08-16 15:49:08 +08:00
|
|
|
err = io.EOF
|
|
|
|
}
|
2020-08-17 06:16:46 +08:00
|
|
|
// fmt.Printf("~~~ filled %d, err: %v\n\n", n, err)
|
2020-03-27 19:50:51 +08:00
|
|
|
|
|
|
|
return
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2022-02-26 19:06:17 +08:00
|
|
|
func (c *ChunkReadAt) readChunkSliceAt(buffer []byte, chunkView *ChunkView, nextChunkViews []*ChunkView, offset uint64) (n int, err error) {
|
2021-04-29 07:13:37 +08:00
|
|
|
|
2021-12-23 08:17:30 +08:00
|
|
|
if c.readerPattern.IsRandomMode() {
|
2022-03-13 17:38:52 +08:00
|
|
|
return fetchChunkRange(buffer, c.readerCache.lookupFileIdFn, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped, int64(offset))
|
2021-04-29 07:13:37 +08:00
|
|
|
}
|
|
|
|
|
2022-02-26 18:16:47 +08:00
|
|
|
n, err = c.readerCache.ReadChunkAt(buffer, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped, int64(offset), int(chunkView.ChunkSize), chunkView.LogicOffset == 0)
|
2022-02-27 19:57:24 +08:00
|
|
|
if c.lastChunkFid != chunkView.FileId {
|
2022-02-27 15:20:45 +08:00
|
|
|
if chunkView.Offset == 0 { // start of a new chunk
|
2022-02-27 19:57:24 +08:00
|
|
|
if c.lastChunkFid != "" {
|
|
|
|
c.readerCache.UnCache(c.lastChunkFid)
|
|
|
|
c.readerCache.MaybeCache(nextChunkViews)
|
|
|
|
} else {
|
|
|
|
if len(nextChunkViews) >= 1 {
|
|
|
|
c.readerCache.MaybeCache(nextChunkViews[:1]) // just read the next chunk if at the very beginning
|
|
|
|
}
|
|
|
|
}
|
2022-02-26 19:06:17 +08:00
|
|
|
}
|
2020-12-09 14:26:46 +08:00
|
|
|
}
|
2022-02-27 15:20:45 +08:00
|
|
|
c.lastChunkFid = chunkView.FileId
|
2020-08-17 06:16:46 +08:00
|
|
|
return
|
2020-03-29 15:54:39 +08:00
|
|
|
}
|