aboutsummaryrefslogtreecommitdiff
path: root/weed/filesys/dirty_page.go
blob: 451bc0d4b0933e18932bc866745cc4e2224de210 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
package filesys

import (
	"bytes"
	"io"
	"runtime"
	"sync"
	"time"

	"github.com/chrislusf/seaweedfs/weed/util/log"
	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
)

type ContinuousDirtyPages struct {
	intervals              *ContinuousIntervals
	f                      *File
	writeWaitGroup         sync.WaitGroup
	chunkAddLock           sync.Mutex
	chunkSaveErrChan       chan error
	chunkSaveErrChanClosed bool
	lastErr                error
	collection             string
	replication            string
}

func newDirtyPages(file *File) *ContinuousDirtyPages {
	dirtyPages := &ContinuousDirtyPages{
		intervals:        &ContinuousIntervals{},
		f:                file,
		chunkSaveErrChan: make(chan error, runtime.NumCPU()),
	}
	go func() {
		for t := range dirtyPages.chunkSaveErrChan {
			if t != nil {
				dirtyPages.lastErr = t
			}
		}
	}()
	return dirtyPages
}

func (pages *ContinuousDirtyPages) AddPage(offset int64, data []byte) {

	log.Tracef("%s AddPage [%d,%d) of %d bytes", pages.f.fullpath(), offset, offset+int64(len(data)), pages.f.entry.Attributes.FileSize)

	if len(data) > int(pages.f.wfs.option.ChunkSizeLimit) {
		// this is more than what buffer can hold.
		pages.flushAndSave(offset, data)
	}

	pages.intervals.AddInterval(data, offset)

	if pages.intervals.TotalSize() >= pages.f.wfs.option.ChunkSizeLimit {
		pages.saveExistingLargestPageToStorage()
	}

	return
}

func (pages *ContinuousDirtyPages) flushAndSave(offset int64, data []byte) {

	// flush existing
	pages.saveExistingPagesToStorage()

	// flush the new page
	pages.saveToStorage(bytes.NewReader(data), offset, int64(len(data)))

	return
}

func (pages *ContinuousDirtyPages) saveExistingPagesToStorage() {
	for pages.saveExistingLargestPageToStorage() {
	}
}

func (pages *ContinuousDirtyPages) saveExistingLargestPageToStorage() (hasSavedData bool) {

	maxList := pages.intervals.RemoveLargestIntervalLinkedList()
	if maxList == nil {
		return false
	}

	fileSize := int64(pages.f.entry.Attributes.FileSize)

	chunkSize := min(maxList.Size(), fileSize-maxList.Offset())
	if chunkSize == 0 {
		return false
	}

	pages.saveToStorage(maxList.ToReader(), maxList.Offset(), chunkSize)

	return true
}

func (pages *ContinuousDirtyPages) saveToStorage(reader io.Reader, offset int64, size int64) {

	errChanSize := pages.f.wfs.option.ConcurrentWriters
	if errChanSize == 0 {
		errChanSize = runtime.NumCPU()
	}
	if pages.chunkSaveErrChanClosed {
		pages.chunkSaveErrChan = make(chan error, errChanSize)
		pages.chunkSaveErrChanClosed = false
	}

	mtime := time.Now().UnixNano()
	pages.writeWaitGroup.Add(1)
	writer := func() {
		defer pages.writeWaitGroup.Done()

		reader = io.LimitReader(reader, size)
		chunk, collection, replication, err := pages.f.wfs.saveDataAsChunk(pages.f.fullpath())(reader, pages.f.Name, offset)
		if err != nil {
			log.Infof("%s saveToStorage [%d,%d): %v", pages.f.fullpath(), offset, offset+size, err)
			pages.chunkSaveErrChan <- err
			return
		}
		chunk.Mtime = mtime
		pages.collection, pages.replication = collection, replication
		pages.chunkAddLock.Lock()
		defer pages.chunkAddLock.Unlock()
		pages.f.addChunks([]*filer_pb.FileChunk{chunk})
		log.Tracef("%s saveToStorage [%d,%d)", pages.f.fullpath(), offset, offset+size)
	}

	if pages.f.wfs.concurrentWriters != nil {
		pages.f.wfs.concurrentWriters.Execute(writer)
	} else {
		go writer()
	}
}

func max(x, y int64) int64 {
	if x > y {
		return x
	}
	return y
}
func min(x, y int64) int64 {
	if x < y {
		return x
	}
	return y
}

func (pages *ContinuousDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
	return pages.intervals.ReadDataAt(data, startOffset)
}