mirror of
https://github.com/seaweedfs/seaweedfs.git
synced 2024-11-24 11:09:12 +08:00
refactor concurrent loading code logic
This commit is contained in:
parent
096ffa9744
commit
b9f385bd66
@ -22,11 +22,43 @@ func NewDiskLocation(dir string, maxVolumeCount int) *DiskLocation {
|
||||
return location
|
||||
}
|
||||
|
||||
func (l *DiskLocation) loadExistingVolumes(needleMapKind NeedleMapType) {
|
||||
l.Lock()
|
||||
defer l.Unlock()
|
||||
func (l *DiskLocation) loadExistingVolume(dir os.FileInfo, needleMapKind NeedleMapType, mutex *sync.RWMutex) {
|
||||
name := dir.Name()
|
||||
if !dir.IsDir() && strings.HasSuffix(name, ".dat") {
|
||||
collection := ""
|
||||
base := name[:len(name)-len(".dat")]
|
||||
i := strings.LastIndex(base, "_")
|
||||
if i > 0 {
|
||||
collection, base = base[0:i], base[i+1:]
|
||||
}
|
||||
if vid, err := NewVolumeId(base); err == nil {
|
||||
mutex.RLock()
|
||||
_, found := l.volumes[vid]
|
||||
mutex.RUnlock()
|
||||
if !found {
|
||||
if v, e := NewVolume(l.Directory, collection, vid, needleMapKind, nil, nil); e == nil {
|
||||
mutex.Lock()
|
||||
l.volumes[vid] = v
|
||||
mutex.Unlock()
|
||||
glog.V(0).Infof("data file %s, replicaPlacement=%s v=%d size=%d ttl=%s", l.Directory+"/"+name, v.ReplicaPlacement, v.Version(), v.Size(), v.Ttl.String())
|
||||
} else {
|
||||
glog.V(0).Infof("new volume %s error %s", name, e)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
task_queue := make(chan os.FileInfo, 100)
|
||||
func (l *DiskLocation) concurrentLoadingVolumes(needleMapKind NeedleMapType, concurrentFlag bool) {
|
||||
var concurrency int
|
||||
if concurrentFlag {
|
||||
//You could choose a better optimized concurency value after testing at your environment
|
||||
concurrency = 10
|
||||
} else {
|
||||
concurrency = 1
|
||||
}
|
||||
|
||||
task_queue := make(chan os.FileInfo, 10*concurrency)
|
||||
go func() {
|
||||
if dirs, err := ioutil.ReadDir(l.Directory); err == nil {
|
||||
for _, dir := range dirs {
|
||||
@ -36,7 +68,6 @@ func (l *DiskLocation) loadExistingVolumes(needleMapKind NeedleMapType) {
|
||||
close(task_queue)
|
||||
}()
|
||||
|
||||
const concurrency int = 10
|
||||
var wg sync.WaitGroup
|
||||
var mutex sync.RWMutex
|
||||
for workerNum := 0; workerNum < concurrency; workerNum++ {
|
||||
@ -44,35 +75,20 @@ func (l *DiskLocation) loadExistingVolumes(needleMapKind NeedleMapType) {
|
||||
go func() {
|
||||
defer wg.Done()
|
||||
for dir := range task_queue {
|
||||
name := dir.Name()
|
||||
if !dir.IsDir() && strings.HasSuffix(name, ".dat") {
|
||||
collection := ""
|
||||
base := name[:len(name)-len(".dat")]
|
||||
i := strings.LastIndex(base, "_")
|
||||
if i > 0 {
|
||||
collection, base = base[0:i], base[i+1:]
|
||||
}
|
||||
if vid, err := NewVolumeId(base); err == nil {
|
||||
mutex.RLock()
|
||||
_, found := l.volumes[vid]
|
||||
mutex.RUnlock()
|
||||
if !found {
|
||||
if v, e := NewVolume(l.Directory, collection, vid, needleMapKind, nil, nil); e == nil {
|
||||
mutex.Lock()
|
||||
l.volumes[vid] = v
|
||||
mutex.Unlock()
|
||||
glog.V(0).Infof("data file %s, replicaPlacement=%s v=%d size=%d ttl=%s", l.Directory+"/"+name, v.ReplicaPlacement, v.Version(), v.Size(), v.Ttl.String())
|
||||
} else {
|
||||
glog.V(0).Infof("new volume %s error %s", name, e)
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
l.loadExistingVolume(dir, needleMapKind, &mutex)
|
||||
}
|
||||
}()
|
||||
}
|
||||
wg.Wait()
|
||||
|
||||
}
|
||||
|
||||
func (l *DiskLocation) loadExistingVolumes(needleMapKind NeedleMapType) {
|
||||
l.Lock()
|
||||
defer l.Unlock()
|
||||
|
||||
l.concurrentLoadingVolumes(needleMapKind, true)
|
||||
|
||||
glog.V(0).Infoln("Store started on dir:", l.Directory, "with", len(l.volumes), "volumes", "max", l.MaxVolumeCount)
|
||||
}
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user