seaweedfs/weed/filer/filechunk_section.go

135 lines
3.8 KiB
Go
Raw Normal View History

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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
package filer
import (
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"sync"
)
2023-01-10 17:35:44 +08:00
const SectionSize = 2 * 1024 * 1024 * 32 // 64MiB
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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
type SectionIndex int64
type FileChunkSection struct {
sectionIndex SectionIndex
chunks []*filer_pb.FileChunk
visibleIntervals *IntervalList[*VisibleInterval]
chunkViews *IntervalList[*ChunkView]
reader *ChunkReadAt
lock sync.Mutex
}
func NewFileChunkSection(si SectionIndex) *FileChunkSection {
return &FileChunkSection{
sectionIndex: si,
}
}
func (section *FileChunkSection) addChunk(chunk *filer_pb.FileChunk) error {
section.lock.Lock()
defer section.lock.Unlock()
start, stop := max(int64(section.sectionIndex)*SectionSize, chunk.Offset), min(((int64(section.sectionIndex)+1)*SectionSize), chunk.Offset+int64(chunk.Size))
section.chunks = append(section.chunks, chunk)
if section.visibleIntervals == nil {
section.visibleIntervals = readResolvedChunks(section.chunks, int64(section.sectionIndex)*SectionSize, (int64(section.sectionIndex)+1)*SectionSize)
} else {
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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
MergeIntoVisibles(section.visibleIntervals, start, stop, chunk)
2023-01-06 17:03:29 +08:00
garbageFileIds := FindGarbageChunks(section.visibleIntervals, start, stop)
2023-01-10 16:46:46 +08:00
removeGarbageChunks(section, garbageFileIds)
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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
}
if section.chunkViews != nil {
MergeIntoChunkViews(section.chunkViews, start, stop, chunk)
}
return nil
}
2023-01-10 16:46:46 +08:00
func removeGarbageChunks(section *FileChunkSection, garbageFileIds map[string]struct{}) {
for i := 0; i < len(section.chunks); {
t := section.chunks[i]
length := len(section.chunks)
if _, found := garbageFileIds[t.FileId]; found {
if i < length-1 {
section.chunks[i] = section.chunks[length-1]
}
section.chunks = section.chunks[:length-1]
} else {
i++
}
}
}
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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
func (section *FileChunkSection) setupForRead(group *ChunkGroup, fileSize int64) {
if section.visibleIntervals == nil {
section.visibleIntervals = readResolvedChunks(section.chunks, int64(section.sectionIndex)*SectionSize, (int64(section.sectionIndex)+1)*SectionSize)
section.chunks, _ = SeparateGarbageChunks(section.visibleIntervals, section.chunks)
if section.reader != nil {
_ = section.reader.Close()
section.reader = nil
}
}
if section.chunkViews == nil {
section.chunkViews = ViewFromVisibleIntervals(section.visibleIntervals, int64(section.sectionIndex)*SectionSize, (int64(section.sectionIndex)+1)*SectionSize)
}
if section.reader == nil {
2023-01-17 14:43:02 +08:00
section.reader = NewChunkReaderAtFromClient(group.readerCache, section.chunkViews, min(int64(section.sectionIndex+1)*SectionSize, fileSize))
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 985137ec472924e4815f258189f6ca9f2168a0a7. * 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
2023-01-03 15:20:45 +08:00
}
section.reader.fileSize = fileSize
}
func (section *FileChunkSection) readDataAt(group *ChunkGroup, fileSize int64, buff []byte, offset int64) (n int, tsNs int64, err error) {
section.lock.Lock()
defer section.lock.Unlock()
section.setupForRead(group, fileSize)
return section.reader.ReadAtWithTime(buff, offset)
}
func (section *FileChunkSection) DataStartOffset(group *ChunkGroup, offset int64, fileSize int64) int64 {
section.lock.Lock()
defer section.lock.Unlock()
section.setupForRead(group, fileSize)
for x := section.visibleIntervals.Front(); x != nil; x = x.Next {
visible := x.Value
if visible.stop <= offset {
continue
}
if offset < visible.start {
return offset
}
return offset
}
return -1
}
func (section *FileChunkSection) NextStopOffset(group *ChunkGroup, offset int64, fileSize int64) int64 {
section.lock.Lock()
defer section.lock.Unlock()
section.setupForRead(group, fileSize)
isAfterOffset := false
for x := section.visibleIntervals.Front(); x != nil; x = x.Next {
visible := x.Value
if !isAfterOffset {
if visible.stop <= offset {
continue
}
isAfterOffset = true
}
if offset < visible.start {
return offset
}
// now visible.start <= offset
if offset < visible.stop {
offset = visible.stop
}
}
return offset
}