seaweedfs/weed/mount/page_writer/upload_pipeline.go

197 lines
6.1 KiB
Go
Raw Normal View History

2022-02-14 06:50:44 +00:00
package page_writer
import (
"fmt"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/util"
2022-02-14 06:50:44 +00:00
"sync"
"sync/atomic"
"time"
)
type LogicChunkIndex int
type UploadPipeline struct {
filepath util.FullPath
ChunkSize int64
writableChunks map[LogicChunkIndex]PageChunk
writableChunksLock sync.Mutex
sealedChunks map[LogicChunkIndex]*SealedChunk
sealedChunksLock sync.Mutex
uploaders *util.LimitedConcurrentExecutor
uploaderCount int32
uploaderCountCond *sync.Cond
saveToStorageFn SaveToStorageFunc
activeReadChunks map[LogicChunkIndex]int
activeReadChunksLock sync.Mutex
2022-03-14 01:34:57 +00:00
writableChunkLimit int
swapFile *SwapFile
2022-02-14 06:50:44 +00:00
}
type SealedChunk struct {
chunk PageChunk
referenceCounter int // track uploading or reading processes
}
func (sc *SealedChunk) FreeReference(messageOnFree string) {
sc.referenceCounter--
if sc.referenceCounter == 0 {
glog.V(4).Infof("Free sealed chunk: %s", messageOnFree)
sc.chunk.FreeResource()
}
}
func NewUploadPipeline(writers *util.LimitedConcurrentExecutor, chunkSize int64, saveToStorageFn SaveToStorageFunc, bufferChunkLimit int, swapFileDir string) *UploadPipeline {
2022-02-14 06:50:44 +00:00
return &UploadPipeline{
2022-03-14 01:34:57 +00:00
ChunkSize: chunkSize,
writableChunks: make(map[LogicChunkIndex]PageChunk),
sealedChunks: make(map[LogicChunkIndex]*SealedChunk),
uploaders: writers,
uploaderCountCond: sync.NewCond(&sync.Mutex{}),
saveToStorageFn: saveToStorageFn,
activeReadChunks: make(map[LogicChunkIndex]int),
writableChunkLimit: bufferChunkLimit,
swapFile: NewSwapFile(swapFileDir, chunkSize),
2022-02-14 06:50:44 +00:00
}
}
func (up *UploadPipeline) SaveDataAt(p []byte, off int64, isSequential bool) (n int) {
2022-02-14 06:50:44 +00:00
up.writableChunksLock.Lock()
defer up.writableChunksLock.Unlock()
logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
pageChunk, found := up.writableChunks[logicChunkIndex]
2022-02-14 06:50:44 +00:00
if !found {
2022-03-14 01:34:57 +00:00
if len(up.writableChunks) > up.writableChunkLimit {
// if current file chunks is over the per file buffer count limit
2022-02-14 06:50:44 +00:00
fullestChunkIndex, fullness := LogicChunkIndex(-1), int64(0)
for lci, mc := range up.writableChunks {
chunkFullness := mc.WrittenSize()
if fullness < chunkFullness {
fullestChunkIndex = lci
fullness = chunkFullness
}
}
up.moveToSealed(up.writableChunks[fullestChunkIndex], fullestChunkIndex)
delete(up.writableChunks, fullestChunkIndex)
2022-03-07 19:22:26 +00:00
// fmt.Printf("flush chunk %d with %d bytes written\n", logicChunkIndex, fullness)
}
if isSequential &&
2022-03-14 01:34:57 +00:00
len(up.writableChunks) < up.writableChunkLimit &&
atomic.LoadInt64(&memChunkCounter) < 4*int64(up.writableChunkLimit) {
pageChunk = NewMemChunk(logicChunkIndex, up.ChunkSize)
} else {
pageChunk = up.swapFile.NewTempFileChunk(logicChunkIndex)
2022-02-14 06:50:44 +00:00
}
up.writableChunks[logicChunkIndex] = pageChunk
2022-02-14 06:50:44 +00:00
}
n = pageChunk.WriteDataAt(p, off)
up.maybeMoveToSealed(pageChunk, logicChunkIndex)
2022-02-14 06:50:44 +00:00
return
}
func (up *UploadPipeline) MaybeReadDataAt(p []byte, off int64) (maxStop int64) {
logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
// read from sealed chunks first
up.sealedChunksLock.Lock()
sealedChunk, found := up.sealedChunks[logicChunkIndex]
if found {
sealedChunk.referenceCounter++
}
up.sealedChunksLock.Unlock()
if found {
maxStop = sealedChunk.chunk.ReadDataAt(p, off)
glog.V(4).Infof("%s read sealed memchunk [%d,%d)", up.filepath, off, maxStop)
sealedChunk.FreeReference(fmt.Sprintf("%s finish reading chunk %d", up.filepath, logicChunkIndex))
}
// read from writable chunks last
up.writableChunksLock.Lock()
defer up.writableChunksLock.Unlock()
writableChunk, found := up.writableChunks[logicChunkIndex]
if !found {
return
}
writableMaxStop := writableChunk.ReadDataAt(p, off)
glog.V(4).Infof("%s read writable memchunk [%d,%d)", up.filepath, off, writableMaxStop)
maxStop = max(maxStop, writableMaxStop)
return
}
func (up *UploadPipeline) FlushAll() {
up.writableChunksLock.Lock()
defer up.writableChunksLock.Unlock()
for logicChunkIndex, memChunk := range up.writableChunks {
up.moveToSealed(memChunk, logicChunkIndex)
}
up.waitForCurrentWritersToComplete()
}
func (up *UploadPipeline) maybeMoveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
if memChunk.IsComplete() {
up.moveToSealed(memChunk, logicChunkIndex)
}
}
func (up *UploadPipeline) moveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
atomic.AddInt32(&up.uploaderCount, 1)
glog.V(4).Infof("%s uploaderCount %d ++> %d", up.filepath, up.uploaderCount-1, up.uploaderCount)
up.sealedChunksLock.Lock()
if oldMemChunk, found := up.sealedChunks[logicChunkIndex]; found {
oldMemChunk.FreeReference(fmt.Sprintf("%s replace chunk %d", up.filepath, logicChunkIndex))
}
sealedChunk := &SealedChunk{
chunk: memChunk,
referenceCounter: 1, // default 1 is for uploading process
}
up.sealedChunks[logicChunkIndex] = sealedChunk
delete(up.writableChunks, logicChunkIndex)
up.sealedChunksLock.Unlock()
up.uploaders.Execute(func() {
// first add to the file chunks
sealedChunk.chunk.SaveContent(up.saveToStorageFn)
// notify waiting process
atomic.AddInt32(&up.uploaderCount, -1)
glog.V(4).Infof("%s uploaderCount %d --> %d", up.filepath, up.uploaderCount+1, up.uploaderCount)
// Lock and Unlock are not required,
// but it may signal multiple times during one wakeup,
// and the waiting goroutine may miss some of them!
up.uploaderCountCond.L.Lock()
up.uploaderCountCond.Broadcast()
up.uploaderCountCond.L.Unlock()
// wait for readers
for up.IsLocked(logicChunkIndex) {
time.Sleep(59 * time.Millisecond)
}
// then remove from sealed chunks
up.sealedChunksLock.Lock()
defer up.sealedChunksLock.Unlock()
delete(up.sealedChunks, logicChunkIndex)
sealedChunk.FreeReference(fmt.Sprintf("%s finished uploading chunk %d", up.filepath, logicChunkIndex))
})
}
func (up *UploadPipeline) Shutdown() {
up.swapFile.FreeResource()
up.sealedChunksLock.Lock()
defer up.sealedChunksLock.Unlock()
2022-03-07 22:01:24 +00:00
for logicChunkIndex, sealedChunk := range up.sealedChunks {
sealedChunk.FreeReference(fmt.Sprintf("%s uploadpipeline shutdown chunk %d", up.filepath, logicChunkIndex))
}
2022-02-14 06:50:44 +00:00
}