seaweedfs/weed/filer/reader_at.go
Chris Lu d4566d4aaa
more solid weed mount (#4089)
* compare chunks by timestamp

* fix slab clearing error

* fix test compilation

* move oldest chunk to sealed, instead of by fullness

* lock on fh.entryViewCache

* remove verbose logs

* revert slat clearing

* less logs

* less logs

* track write and read by timestamp

* remove useless logic

* add entry lock on file handle release

* use mem chunk only, swap file chunk has problems

* comment out code that maybe used later

* add debug mode to compare data read and write

* more efficient readResolvedChunks with linked list

* small optimization

* fix test compilation

* minor fix on writer

* add SeparateGarbageChunks

* group chunks into sections

* turn off debug mode

* fix tests

* fix tests

* tmp enable swap file chunk

* Revert "tmp enable swap file chunk"

This reverts commit 985137ec47.

* simple refactoring

* simple refactoring

* do not re-use swap file chunk. Sealed chunks should not be re-used.

* comment out debugging facilities

* either mem chunk or swap file chunk is fine now

* remove orderedMutex  as *semaphore.Weighted

not found impactful

* optimize size calculation for changing large files

* optimize performance to avoid going through the long list of chunks

* still problems with swap file chunk

* rename

* tiny optimization

* swap file chunk save only successfully read data

* fix

* enable both mem and swap file chunk

* resolve chunks with range

* rename

* fix chunk interval list

* also change file handle chunk group when adding chunks

* pick in-active chunk with time-decayed counter

* fix compilation

* avoid nil with empty fh.entry

* refactoring

* rename

* rename

* refactor visible intervals to *list.List

* refactor chunkViews to *list.List

* add IntervalList for generic interval list

* change visible interval to use IntervalList in generics

* cahnge chunkViews to *IntervalList[*ChunkView]

* use NewFileChunkSection to create

* rename variables

* refactor

* fix renaming leftover

* renaming

* renaming

* add insert interval

* interval list adds lock

* incrementally add chunks to readers

Fixes:
1. set start and stop offset for the value object
2. clone the value object
3. use pointer instead of copy-by-value when passing to interval.Value
4. use insert interval since adding chunk could be out of order

* fix tests compilation

* fix tests compilation
2023-01-02 23:20:45 -08:00

228 lines
7.1 KiB
Go

package filer
import (
"context"
"fmt"
"io"
"math/rand"
"sync"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/util"
"github.com/seaweedfs/seaweedfs/weed/util/chunk_cache"
"github.com/seaweedfs/seaweedfs/weed/wdclient"
)
type ChunkReadAt struct {
masterClient *wdclient.MasterClient
chunkViews *IntervalList[*ChunkView]
fileSize int64
readerCache *ReaderCache
readerPattern *ReaderPattern
lastChunkFid string
}
var _ = io.ReaderAt(&ChunkReadAt{})
var _ = io.Closer(&ChunkReadAt{})
func LookupFn(filerClient filer_pb.FilerClient) wdclient.LookupFileIdFunctionType {
vidCache := make(map[string]*filer_pb.Locations)
var vicCacheLock sync.RWMutex
return func(fileId string) (targetUrls []string, err error) {
vid := VolumeId(fileId)
vicCacheLock.RLock()
locations, found := vidCache[vid]
vicCacheLock.RUnlock()
if !found {
util.Retry("lookup volume "+vid, func() error {
err = filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
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
})
return err
})
}
if err != nil {
return nil, err
}
fcDataCenter := filerClient.GetDataCenter()
var sameDcTargetUrls, otherTargetUrls []string
for _, loc := range locations.Locations {
volumeServerAddress := filerClient.AdjustedUrl(loc)
targetUrl := fmt.Sprintf("http://%s/%s", volumeServerAddress, fileId)
if fcDataCenter == "" || fcDataCenter != loc.DataCenter {
otherTargetUrls = append(otherTargetUrls, targetUrl)
} else {
sameDcTargetUrls = append(sameDcTargetUrls, targetUrl)
}
}
rand.Shuffle(len(sameDcTargetUrls), func(i, j int) {
sameDcTargetUrls[i], sameDcTargetUrls[j] = sameDcTargetUrls[j], sameDcTargetUrls[i]
})
rand.Shuffle(len(otherTargetUrls), func(i, j int) {
otherTargetUrls[i], otherTargetUrls[j] = otherTargetUrls[j], otherTargetUrls[i]
})
// Prefer same data center
targetUrls = append(sameDcTargetUrls, otherTargetUrls...)
return
}
}
func NewChunkReaderAtFromClient(lookupFn wdclient.LookupFileIdFunctionType, chunkViews *IntervalList[*ChunkView], chunkCache chunk_cache.ChunkCache, fileSize int64) *ChunkReadAt {
return &ChunkReadAt{
chunkViews: chunkViews,
fileSize: fileSize,
readerCache: newReaderCache(32, chunkCache, lookupFn),
readerPattern: NewReaderPattern(),
}
}
func (c *ChunkReadAt) Close() error {
c.readerCache.destroy()
return nil
}
func (c *ChunkReadAt) ReadAt(p []byte, offset int64) (n int, err error) {
c.readerPattern.MonitorReadAt(offset, len(p))
c.chunkViews.Lock.Lock()
defer c.chunkViews.Lock.Unlock()
// 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))
n, _, err = c.doReadAt(p, offset)
return
}
func (c *ChunkReadAt) ReadAtWithTime(p []byte, offset int64) (n int, ts int64, err error) {
c.readerPattern.MonitorReadAt(offset, len(p))
c.chunkViews.Lock.Lock()
defer c.chunkViews.Lock.Unlock()
// 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))
return c.doReadAt(p, offset)
}
func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, ts int64, err error) {
startOffset, remaining := offset, int64(len(p))
var nextChunks *Interval[*ChunkView]
for x := c.chunkViews.Front(); x != nil; x = x.Next {
chunk := x.Value
if remaining <= 0 {
break
}
if x.Next != nil {
nextChunks = x.Next
}
if startOffset < chunk.ViewOffset {
gap := chunk.ViewOffset - startOffset
glog.V(4).Infof("zero [%d,%d)", startOffset, chunk.ViewOffset)
n += zero(p, startOffset-offset, gap)
startOffset, remaining = chunk.ViewOffset, remaining-gap
if remaining <= 0 {
break
}
}
// fmt.Printf(">>> doReadAt [%d,%d), chunk[%d,%d)\n", offset, offset+int64(len(p)), chunk.ViewOffset, chunk.ViewOffset+int64(chunk.ViewSize))
chunkStart, chunkStop := max(chunk.ViewOffset, startOffset), min(chunk.ViewOffset+int64(chunk.ViewSize), startOffset+remaining)
if chunkStart >= chunkStop {
continue
}
// glog.V(4).Infof("read [%d,%d), %d/%d chunk %s [%d,%d)", chunkStart, chunkStop, i, len(c.chunkViews), chunk.FileId, chunk.ViewOffset-chunk.Offset, chunk.ViewOffset-chunk.Offset+int64(chunk.ViewSize))
bufferOffset := chunkStart - chunk.ViewOffset + chunk.OffsetInChunk
ts = chunk.ModifiedTsNs
copied, err := c.readChunkSliceAt(p[startOffset-offset:chunkStop-chunkStart+startOffset-offset], chunk, nextChunks, uint64(bufferOffset))
if err != nil {
glog.Errorf("fetching chunk %+v: %v\n", chunk, err)
return copied, ts, err
}
n += copied
startOffset, remaining = startOffset+int64(copied), remaining-int64(copied)
}
// glog.V(4).Infof("doReadAt [%d,%d), n:%v, err:%v", offset, offset+int64(len(p)), n, err)
// zero the remaining bytes if a gap exists at the end of the last chunk (or a fully sparse file)
if err == nil && remaining > 0 {
var delta int64
if c.fileSize > startOffset {
delta = min(remaining, c.fileSize-startOffset)
startOffset -= offset
} else {
delta = remaining
startOffset = max(startOffset-offset, startOffset-remaining-offset)
}
glog.V(4).Infof("zero2 [%d,%d) of file size %d bytes", startOffset, startOffset+delta, c.fileSize)
n += zero(p, startOffset, delta)
}
if err == nil && offset+int64(len(p)) >= c.fileSize {
err = io.EOF
}
// fmt.Printf("~~~ filled %d, err: %v\n\n", n, err)
return
}
func (c *ChunkReadAt) readChunkSliceAt(buffer []byte, chunkView *ChunkView, nextChunkViews *Interval[*ChunkView], offset uint64) (n int, err error) {
if c.readerPattern.IsRandomMode() {
n, err := c.readerCache.chunkCache.ReadChunkAt(buffer, chunkView.FileId, offset)
if n > 0 {
return n, err
}
return fetchChunkRange(buffer, c.readerCache.lookupFileIdFn, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped, int64(offset))
}
n, err = c.readerCache.ReadChunkAt(buffer, chunkView.FileId, chunkView.CipherKey, chunkView.IsGzipped, int64(offset), int(chunkView.ChunkSize), chunkView.ViewOffset == 0)
if c.lastChunkFid != chunkView.FileId {
if chunkView.OffsetInChunk == 0 { // start of a new chunk
if c.lastChunkFid != "" {
c.readerCache.UnCache(c.lastChunkFid)
}
if nextChunkViews != nil {
c.readerCache.MaybeCache(nextChunkViews) // just read the next chunk if at the very beginning
}
}
}
c.lastChunkFid = chunkView.FileId
return
}
func zero(buffer []byte, start, length int64) int {
end := min(start+length, int64(len(buffer)))
start = max(start, 0)
// zero the bytes
for o := start; o < end; o++ {
buffer[o] = 0
}
return int(end - start)
}