mirror of
https://github.com/seaweedfs/seaweedfs.git
synced 2024-11-30 23:29:02 +08:00
341 lines
9.5 KiB
Go
341 lines
9.5 KiB
Go
package topology
|
|
|
|
import (
|
|
"encoding/json"
|
|
"errors"
|
|
"fmt"
|
|
"math/rand"
|
|
"sync"
|
|
"time"
|
|
|
|
"github.com/seaweedfs/seaweedfs/weed/pb"
|
|
"github.com/seaweedfs/seaweedfs/weed/storage/types"
|
|
|
|
hashicorpRaft "github.com/hashicorp/raft"
|
|
"github.com/seaweedfs/raft"
|
|
|
|
"github.com/seaweedfs/seaweedfs/weed/glog"
|
|
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
|
|
"github.com/seaweedfs/seaweedfs/weed/sequence"
|
|
"github.com/seaweedfs/seaweedfs/weed/storage"
|
|
"github.com/seaweedfs/seaweedfs/weed/storage/needle"
|
|
"github.com/seaweedfs/seaweedfs/weed/storage/super_block"
|
|
"github.com/seaweedfs/seaweedfs/weed/util"
|
|
)
|
|
|
|
type Topology struct {
|
|
vacuumLockCounter int64
|
|
NodeImpl
|
|
|
|
collectionMap *util.ConcurrentReadMap
|
|
ecShardMap map[needle.VolumeId]*EcShardLocations
|
|
ecShardMapLock sync.RWMutex
|
|
|
|
pulse int64
|
|
|
|
volumeSizeLimit uint64
|
|
replicationAsMin bool
|
|
|
|
Sequence sequence.Sequencer
|
|
|
|
chanFullVolumes chan storage.VolumeInfo
|
|
chanCrowdedVolumes chan storage.VolumeInfo
|
|
|
|
Configuration *Configuration
|
|
|
|
RaftServer raft.Server
|
|
RaftServerAccessLock sync.RWMutex
|
|
HashicorpRaft *hashicorpRaft.Raft
|
|
UuidAccessLock sync.RWMutex
|
|
UuidMap map[string][]string
|
|
}
|
|
|
|
func NewTopology(id string, seq sequence.Sequencer, volumeSizeLimit uint64, pulse int, replicationAsMin bool) *Topology {
|
|
t := &Topology{}
|
|
t.id = NodeId(id)
|
|
t.nodeType = "Topology"
|
|
t.NodeImpl.value = t
|
|
t.diskUsages = newDiskUsages()
|
|
t.children = make(map[NodeId]Node)
|
|
t.collectionMap = util.NewConcurrentReadMap()
|
|
t.ecShardMap = make(map[needle.VolumeId]*EcShardLocations)
|
|
t.pulse = int64(pulse)
|
|
t.volumeSizeLimit = volumeSizeLimit
|
|
t.replicationAsMin = replicationAsMin
|
|
|
|
t.Sequence = seq
|
|
|
|
t.chanFullVolumes = make(chan storage.VolumeInfo)
|
|
t.chanCrowdedVolumes = make(chan storage.VolumeInfo)
|
|
|
|
t.Configuration = &Configuration{}
|
|
|
|
return t
|
|
}
|
|
|
|
func (t *Topology) IsLeader() bool {
|
|
t.RaftServerAccessLock.RLock()
|
|
defer t.RaftServerAccessLock.RUnlock()
|
|
|
|
if t.RaftServer != nil {
|
|
if t.RaftServer.State() == raft.Leader {
|
|
return true
|
|
}
|
|
if leader, err := t.Leader(); err == nil {
|
|
if pb.ServerAddress(t.RaftServer.Name()) == leader {
|
|
return true
|
|
}
|
|
}
|
|
} else if t.HashicorpRaft != nil {
|
|
if t.HashicorpRaft.State() == hashicorpRaft.Leader {
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
}
|
|
|
|
func (t *Topology) Leader() (l pb.ServerAddress, err error) {
|
|
for count := 0; count < 3; count++ {
|
|
l, err = t.MaybeLeader()
|
|
if err != nil {
|
|
return
|
|
}
|
|
if l != "" {
|
|
break
|
|
}
|
|
|
|
time.Sleep(time.Duration(5+count) * time.Second)
|
|
}
|
|
|
|
return
|
|
}
|
|
|
|
func (t *Topology) MaybeLeader() (l pb.ServerAddress, err error) {
|
|
t.RaftServerAccessLock.RLock()
|
|
defer t.RaftServerAccessLock.RUnlock()
|
|
|
|
if t.RaftServer != nil {
|
|
l = pb.ServerAddress(t.RaftServer.Leader())
|
|
} else if t.HashicorpRaft != nil {
|
|
l = pb.ServerAddress(t.HashicorpRaft.Leader())
|
|
} else {
|
|
err = errors.New("Raft Server not ready yet!")
|
|
}
|
|
|
|
return
|
|
}
|
|
|
|
func (t *Topology) Lookup(collection string, vid needle.VolumeId) (dataNodes []*DataNode) {
|
|
// maybe an issue if lots of collections?
|
|
if collection == "" {
|
|
for _, c := range t.collectionMap.Items() {
|
|
if list := c.(*Collection).Lookup(vid); list != nil {
|
|
return list
|
|
}
|
|
}
|
|
} else {
|
|
if c, ok := t.collectionMap.Find(collection); ok {
|
|
return c.(*Collection).Lookup(vid)
|
|
}
|
|
}
|
|
|
|
if locations, found := t.LookupEcShards(vid); found {
|
|
for _, loc := range locations.Locations {
|
|
dataNodes = append(dataNodes, loc...)
|
|
}
|
|
return dataNodes
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
func (t *Topology) NextVolumeId() (needle.VolumeId, error) {
|
|
vid := t.GetMaxVolumeId()
|
|
next := vid.Next()
|
|
|
|
t.RaftServerAccessLock.RLock()
|
|
defer t.RaftServerAccessLock.RUnlock()
|
|
|
|
if t.RaftServer != nil {
|
|
if _, err := t.RaftServer.Do(NewMaxVolumeIdCommand(next)); err != nil {
|
|
return 0, err
|
|
}
|
|
} else if t.HashicorpRaft != nil {
|
|
b, err := json.Marshal(NewMaxVolumeIdCommand(next))
|
|
if err != nil {
|
|
return 0, fmt.Errorf("failed marshal NewMaxVolumeIdCommand: %+v", err)
|
|
}
|
|
if future := t.HashicorpRaft.Apply(b, time.Second); future.Error() != nil {
|
|
return 0, future.Error()
|
|
}
|
|
}
|
|
return next, nil
|
|
}
|
|
|
|
// deprecated
|
|
func (t *Topology) HasWritableVolume(option *VolumeGrowOption) bool {
|
|
vl := t.GetVolumeLayout(option.Collection, option.ReplicaPlacement, option.Ttl, option.DiskType)
|
|
active, _ := vl.GetActiveVolumeCount(option)
|
|
return active > 0
|
|
}
|
|
|
|
func (t *Topology) PickForWrite(count uint64, option *VolumeGrowOption) (string, uint64, *VolumeLocationList, error) {
|
|
vid, count, datanodes, err := t.GetVolumeLayout(option.Collection, option.ReplicaPlacement, option.Ttl, option.DiskType).PickForWrite(count, option)
|
|
if err != nil {
|
|
return "", 0, nil, fmt.Errorf("failed to find writable volumes for collection:%s replication:%s ttl:%s error: %v", option.Collection, option.ReplicaPlacement.String(), option.Ttl.String(), err)
|
|
}
|
|
if datanodes.Length() == 0 {
|
|
return "", 0, nil, fmt.Errorf("no writable volumes available for collection:%s replication:%s ttl:%s", option.Collection, option.ReplicaPlacement.String(), option.Ttl.String())
|
|
}
|
|
fileId := t.Sequence.NextFileId(count)
|
|
return needle.NewFileId(*vid, fileId, rand.Uint32()).String(), count, datanodes, nil
|
|
}
|
|
|
|
func (t *Topology) GetVolumeLayout(collectionName string, rp *super_block.ReplicaPlacement, ttl *needle.TTL, diskType types.DiskType) *VolumeLayout {
|
|
return t.collectionMap.Get(collectionName, func() interface{} {
|
|
return NewCollection(collectionName, t.volumeSizeLimit, t.replicationAsMin)
|
|
}).(*Collection).GetOrCreateVolumeLayout(rp, ttl, diskType)
|
|
}
|
|
|
|
func (t *Topology) ListCollections(includeNormalVolumes, includeEcVolumes bool) (ret []string) {
|
|
|
|
mapOfCollections := make(map[string]bool)
|
|
for _, c := range t.collectionMap.Items() {
|
|
mapOfCollections[c.(*Collection).Name] = true
|
|
}
|
|
|
|
if includeEcVolumes {
|
|
t.ecShardMapLock.RLock()
|
|
for _, ecVolumeLocation := range t.ecShardMap {
|
|
mapOfCollections[ecVolumeLocation.Collection] = true
|
|
}
|
|
t.ecShardMapLock.RUnlock()
|
|
}
|
|
|
|
for k := range mapOfCollections {
|
|
ret = append(ret, k)
|
|
}
|
|
return ret
|
|
}
|
|
|
|
func (t *Topology) FindCollection(collectionName string) (*Collection, bool) {
|
|
c, hasCollection := t.collectionMap.Find(collectionName)
|
|
if !hasCollection {
|
|
return nil, false
|
|
}
|
|
return c.(*Collection), hasCollection
|
|
}
|
|
|
|
func (t *Topology) DeleteCollection(collectionName string) {
|
|
t.collectionMap.Delete(collectionName)
|
|
}
|
|
|
|
func (t *Topology) DeleteLayout(collectionName string, rp *super_block.ReplicaPlacement, ttl *needle.TTL, diskType types.DiskType) {
|
|
collection, found := t.FindCollection(collectionName)
|
|
if !found {
|
|
return
|
|
}
|
|
collection.DeleteVolumeLayout(rp, ttl, diskType)
|
|
if len(collection.storageType2VolumeLayout.Items()) == 0 {
|
|
t.DeleteCollection(collectionName)
|
|
}
|
|
}
|
|
|
|
func (t *Topology) RegisterVolumeLayout(v storage.VolumeInfo, dn *DataNode) {
|
|
diskType := types.ToDiskType(v.DiskType)
|
|
vl := t.GetVolumeLayout(v.Collection, v.ReplicaPlacement, v.Ttl, diskType)
|
|
vl.RegisterVolume(&v, dn)
|
|
vl.EnsureCorrectWritables(&v)
|
|
}
|
|
func (t *Topology) UnRegisterVolumeLayout(v storage.VolumeInfo, dn *DataNode) {
|
|
glog.Infof("removing volume info: %+v from %v", v, dn.id)
|
|
diskType := types.ToDiskType(v.DiskType)
|
|
volumeLayout := t.GetVolumeLayout(v.Collection, v.ReplicaPlacement, v.Ttl, diskType)
|
|
volumeLayout.UnRegisterVolume(&v, dn)
|
|
if volumeLayout.isEmpty() {
|
|
t.DeleteLayout(v.Collection, v.ReplicaPlacement, v.Ttl, diskType)
|
|
}
|
|
}
|
|
|
|
func (t *Topology) GetOrCreateDataCenter(dcName string) *DataCenter {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
for _, c := range t.children {
|
|
dc := c.(*DataCenter)
|
|
if string(dc.Id()) == dcName {
|
|
return dc
|
|
}
|
|
}
|
|
dc := NewDataCenter(dcName)
|
|
t.doLinkChildNode(dc)
|
|
return dc
|
|
}
|
|
|
|
func (t *Topology) SyncDataNodeRegistration(volumes []*master_pb.VolumeInformationMessage, dn *DataNode) (newVolumes, deletedVolumes []storage.VolumeInfo) {
|
|
// convert into in memory struct storage.VolumeInfo
|
|
var volumeInfos []storage.VolumeInfo
|
|
for _, v := range volumes {
|
|
if vi, err := storage.NewVolumeInfo(v); err == nil {
|
|
volumeInfos = append(volumeInfos, vi)
|
|
} else {
|
|
glog.V(0).Infof("Fail to convert joined volume information: %v", err)
|
|
}
|
|
}
|
|
// find out the delta volumes
|
|
var changedVolumes []storage.VolumeInfo
|
|
newVolumes, deletedVolumes, changedVolumes = dn.UpdateVolumes(volumeInfos)
|
|
for _, v := range newVolumes {
|
|
t.RegisterVolumeLayout(v, dn)
|
|
}
|
|
for _, v := range deletedVolumes {
|
|
t.UnRegisterVolumeLayout(v, dn)
|
|
}
|
|
for _, v := range changedVolumes {
|
|
diskType := types.ToDiskType(v.DiskType)
|
|
vl := t.GetVolumeLayout(v.Collection, v.ReplicaPlacement, v.Ttl, diskType)
|
|
vl.EnsureCorrectWritables(&v)
|
|
}
|
|
return
|
|
}
|
|
|
|
func (t *Topology) IncrementalSyncDataNodeRegistration(newVolumes, deletedVolumes []*master_pb.VolumeShortInformationMessage, dn *DataNode) {
|
|
var newVis, oldVis []storage.VolumeInfo
|
|
for _, v := range newVolumes {
|
|
vi, err := storage.NewVolumeInfoFromShort(v)
|
|
if err != nil {
|
|
glog.V(0).Infof("NewVolumeInfoFromShort %v: %v", v, err)
|
|
continue
|
|
}
|
|
newVis = append(newVis, vi)
|
|
}
|
|
for _, v := range deletedVolumes {
|
|
vi, err := storage.NewVolumeInfoFromShort(v)
|
|
if err != nil {
|
|
glog.V(0).Infof("NewVolumeInfoFromShort %v: %v", v, err)
|
|
continue
|
|
}
|
|
oldVis = append(oldVis, vi)
|
|
}
|
|
dn.DeltaUpdateVolumes(newVis, oldVis)
|
|
|
|
for _, vi := range newVis {
|
|
t.RegisterVolumeLayout(vi, dn)
|
|
}
|
|
for _, vi := range oldVis {
|
|
t.UnRegisterVolumeLayout(vi, dn)
|
|
}
|
|
|
|
return
|
|
}
|
|
|
|
func (t *Topology) DataNodeRegistration(dcName, rackName string, dn *DataNode) {
|
|
if dn.Parent() != nil {
|
|
return
|
|
}
|
|
// registration to topo
|
|
dc := t.GetOrCreateDataCenter(dcName)
|
|
rack := dc.GetOrCreateRack(rackName)
|
|
rack.LinkChildNode(dn)
|
|
glog.Infof("[%s] reLink To topo ", dn.Id())
|
|
}
|