Documentation ¶
Index ¶
- Constants
- Variables
- func DeleteLimiterWait()
- func GetIoMetricLabels(partition *DataPartition, tp string) map[string]string
- func IsDiskErr(errMsg string) bool
- func MarshalRandWriteRaftLog(opcode uint8, extentID uint64, offset, size int64, data []byte, crc uint32) (result []byte, err error)
- func NewPacketToBroadcastMinAppliedID(partitionID uint64, minAppliedID uint64) (p *repl.Packet)
- func NewPacketToGetAppliedID(partitionID uint64) (p *repl.Packet)
- func NewPacketToGetMaxExtentIDAndPartitionSIze(partitionID uint64) (p *repl.Packet)
- func NewPacketToGetPartitionSize(partitionID uint64) (p *repl.Packet)
- func UnmarshalOldVersionRaftLog(raw []byte) (opItem *rndWrtOpItem, err error)
- func UnmarshalOldVersionRandWriteOpItem(raw []byte) (result *rndWrtOpItem, err error)
- func UnmarshalRandWriteRaftLog(raw []byte) (opItem *rndWrtOpItem, err error)
- type DataNode
- type DataNodeInfo
- type DataNodeMetrics
- type DataPartition
- func (dp *DataPartition) Apply(command []byte, index uint64) (resp interface{}, err error)
- func (dp *DataPartition) ApplyMemberChange(confChange *raftproto.ConfChange, index uint64) (resp interface{}, err error)
- func (dp *DataPartition) ApplyRandomWrite(command []byte, raftApplyID uint64) (resp interface{}, err error)
- func (dp *DataPartition) ApplySnapshot(peers []raftproto.Peer, iterator raftproto.SnapIterator) (err error)
- func (dp *DataPartition) Available() int
- func (dp *DataPartition) CanRemoveRaftMember(peer proto.Peer) error
- func (dp *DataPartition) ChangeRaftMember(changeType raftProto.ConfChangeType, peer raftProto.Peer, context []byte) (resp interface{}, err error)
- func (dp *DataPartition) CheckLeader(request *repl.Packet, connect net.Conn) (err error)
- func (dp *DataPartition) Del(key interface{}) (interface{}, error)
- func (dp *DataPartition) Disk() *Disk
- func (dp *DataPartition) DoExtentStoreRepair(repairTask *DataPartitionRepairTask)
- func (dp *DataPartition) DoRepair(repairTasks []*DataPartitionRepairTask)
- func (dp *DataPartition) ExtentStore() *storage.ExtentStore
- func (dp *DataPartition) ForceLoadHeader()
- func (dp *DataPartition) ForceSetDataPartitionToFininshLoad()
- func (dp *DataPartition) ForceSetDataPartitionToLoadding()
- func (dp *DataPartition) ForceSetRaftRunning()
- func (dp *DataPartition) Get(key interface{}) (interface{}, error)
- func (dp *DataPartition) GetAppliedID() (id uint64)
- func (dp *DataPartition) GetExtentCount() int
- func (dp *DataPartition) HandleFatalEvent(err *raft.FatalError)
- func (dp *DataPartition) HandleLeaderChange(leader uint64)
- func (dp *DataPartition) IsEquareCreateDataPartitionRequst(request *proto.CreateDataPartitionRequest) (err error)
- func (dp *DataPartition) IsExsitReplica(addr string) bool
- func (dp *DataPartition) IsRaftLeader() (addr string, ok bool)
- func (dp *DataPartition) IsRejectWrite() bool
- func (dp *DataPartition) LaunchRepair(extentType uint8)
- func (dp *DataPartition) Load() (response *proto.LoadDataPartitionResponse)
- func (dp *DataPartition) LoadAppliedID() (err error)
- func (dp *DataPartition) NotifyExtentRepair(members []*DataPartitionRepairTask) (err error)
- func (dp *DataPartition) Path() string
- func (dp *DataPartition) PersistMetadata() (err error)
- func (dp *DataPartition) Put(key interface{}, val interface{}) (resp interface{}, err error)
- func (dp *DataPartition) RandomWriteSubmit(pkg *repl.Packet) (err error)
- func (dp *DataPartition) ReloadSnapshot()
- func (dp *DataPartition) Replicas() []string
- func (dp *DataPartition) SetMinAppliedID(id uint64)
- func (dp *DataPartition) Size() int
- func (dp *DataPartition) SnapShot() (files []*proto.File)
- func (dp *DataPartition) Snapshot() (raftproto.Snapshot, error)
- func (dp *DataPartition) StartRaft() (err error)
- func (dp *DataPartition) StartRaftAfterRepair()
- func (dp *DataPartition) StartRaftLoggingSchedule()
- func (dp *DataPartition) Status() int
- func (dp *DataPartition) Stop()
- func (dp *DataPartition) String() (m string)
- func (dp *DataPartition) Used() int
- type DataPartitionMetadata
- type DataPartitionRepairTask
- type Disk
- func (d *Disk) AddSize(size uint64)
- func (d *Disk) AttachDataPartition(dp *DataPartition)
- func (d *Disk) DataPartitionList() (partitionIDs []uint64)
- func (d *Disk) DetachDataPartition(dp *DataPartition)
- func (d *Disk) ForceExitRaftStore()
- func (d *Disk) GetDataPartition(partitionID uint64) (partition *DataPartition)
- func (d *Disk) PartitionCount() int
- func (d *Disk) RestorePartition(visitor PartitionVisitor)
- type ItemIterator
- type PartitionVisitor
- type RaftCmdItem
- type SimpleVolView
- type SpaceManager
- func (manager *SpaceManager) AttachPartition(dp *DataPartition)
- func (manager *SpaceManager) CreatePartition(request *proto.CreateDataPartitionRequest) (dp *DataPartition, err error)
- func (manager *SpaceManager) DeletePartition(dpID uint64)
- func (manager *SpaceManager) DetachDataPartition(partitionID uint64)
- func (manager *SpaceManager) GetClusterID() (clusterID string)
- func (manager *SpaceManager) GetDisk(path string) (d *Disk, err error)
- func (manager *SpaceManager) GetDisks() (disks []*Disk)
- func (manager *SpaceManager) GetNodeID() (nodeID uint64)
- func (manager *SpaceManager) GetRaftStore() (raftStore raftstore.RaftStore)
- func (manager *SpaceManager) LoadDisk(path string, reservedSpace uint64, maxErrCnt int) (err error)
- func (manager *SpaceManager) Partition(partitionID uint64) (dp *DataPartition)
- func (manager *SpaceManager) RangePartitions(f func(partition *DataPartition) bool)
- func (manager *SpaceManager) SetClusterID(clusterID string)
- func (manager *SpaceManager) SetNodeID(nodeID uint64)
- func (manager *SpaceManager) SetRaftStore(raftStore raftstore.RaftStore)
- func (manager *SpaceManager) Stats() *Stats
- func (manager *SpaceManager) Stop()
- type Stats
- type VolMap
Constants ¶
const ( IntervalToUpdateReplica = 600 // interval to update the replica IntervalToUpdatePartitionSize = 60 // interval to update the partition size NumOfFilesToRecoverInParallel = 10 // number of files to be recovered simultaneously )
const ( ReadFlag = 1 WriteFlag = 2 )
cmd response
const ( ActionNotifyFollowerToRepair = "ActionNotifyFollowerRepair" ActionStreamRead = "ActionStreamRead" ActionCreateExtent = "ActionCreateExtent:" ActionMarkDelete = "ActionMarkDelete:" ActionGetAllExtentWatermarks = "ActionGetAllExtentWatermarks:" ActionWrite = "ActionWrite:" ActionRepair = "ActionRepair:" ActionDecommissionPartition = "ActionDecommissionPartition" ActionAddDataPartitionRaftMember = "ActionAddDataPartitionRaftMember" ActionRemoveDataPartitionRaftMember = "ActionRemoveDataPartitionRaftMember" ActionDataPartitionTryToLeader = "ActionDataPartitionTryToLeader" ActionCreateDataPartition = "ActionCreateDataPartition" ActionLoadDataPartition = "ActionLoadDataPartition" ActionDeleteDataPartition = "ActionDeleteDataPartition" ActionStreamReadTinyDeleteRecord = "ActionStreamReadTinyDeleteRecord" ActionSyncTinyDeleteRecord = "ActionSyncTinyDeleteRecord" ActionStreamReadTinyExtentRepair = "ActionStreamReadTinyExtentRepair" ActionBatchMarkDelete = "ActionBatchMarkDelete" )
Action description
const ( RepairRead = true StreamRead = false )
const ( EmptyResponse = 'E' TinyExtentRepairReadResponseArgLen = 17 MaxSyncTinyDeleteBufferSize = 2400000 MaxFullSyncTinyDeleteTime = 3600 * 24 MinTinyExtentDeleteRecordSyncSize = 4 * 1024 * 1024 )
const ( MetricPartitionIOName = "dataPartitionIO" MetricPartitionIOBytesName = "dataPartitionIOBytes" )
const ( DataPartitionPrefix = "datapartition" CachePartitionPrefix = "cachepartition" PreLoadPartitionPrefix = "preloadpartition" DataPartitionMetadataFileName = "META" TempMetadataFileName = ".meta" ApplyIndexFile = "APPLY" TempApplyIndexFile = ".apply" TimeLayout = "2006-01-02 15:04:05" )
const ( RaftStatusStopped = 0 RaftStatusRunning = 1 )
const ( DefaultZoneName = proto.DefaultZoneName DefaultRaftDir = "raft" DefaultRaftLogsToRetain = 10 // Count of raft logs per data partition DefaultDiskMaxErr = 1 DefaultDiskRetainMin = 5 * util.GB // GB DefaultDiskRetainMax = 30 * util.GB // GB )
const ( ConfigKeyLocalIP = "localIP" // string ConfigKeyPort = "port" // int ConfigKeyMasterAddr = "masterAddr" // array ConfigKeyZone = "zoneName" // string ConfigKeyDisks = "disks" // array ConfigKeyRaftDir = "raftDir" // string ConfigKeyRaftHeartbeat = "raftHeartbeat" // string ConfigKeyRaftReplica = "raftReplica" // string CfgTickInterval = "tickInterval" // int // smux Config ConfigKeyEnableSmuxClient = "enableSmuxConnPool" //bool ConfigKeySmuxPortShift = "smuxPortShift" //int ConfigKeySmuxMaxConn = "smuxMaxConn" //int ConfigKeySmuxStreamPerConn = "smuxStreamPerConn" //int ConfigKeySmuxMaxBuffer = "smuxMaxBuffer" //int )
const (
BinaryMarshalMagicVersion = 0xFF
)
const (
BufferWrite = false
)
const (
DiskSectorSize = 512
)
Sector size
const (
DiskStatusFile = ".diskStatus"
)
const ExpiredPartitionPrefix = "expired_"
const (
FinishLoadDataPartitionExtentHeader = 1
)
Status of load data partition extent header
const (
IsReleased = 1
)
Tiny extent has been put back to store
const (
MinAvaliTinyExtentCnt = 5
)
const (
MinTinyExtentsToRepair = 10 // minimum number of tiny extents to repair
)
Apply the raft log operation. Currently we only have the random write operation.
const (
ModuleName = "dataNode"
)
const (
NetworkProtocol = "tcp"
)
Network protocol
const (
RaftNotStarted = "RaftNotStarted"
)
Error code
const (
SyncTinyDeleteRecordFromLeaderOnEveryDisk = 5
)
const (
UpdateNodeInfoTicket = 1 * time.Minute
)
Variables ¶
var ( // RegexpDataPartitionDir validates the directory name of a data partition. RegexpDataPartitionDir, _ = regexp.Compile("^datapartition_(\\d)+_(\\d)+$") RegexpCachePartitionDir, _ = regexp.Compile("^cachepartition_(\\d)+_(\\d)+$") RegexpPreLoadPartitionDir, _ = regexp.Compile("^preloadpartition_(\\d)+_(\\d)+$") )
var ( MaxExtentRepairLimit = 20000 MinExtentRepairLimit = 5 CurExtentRepairLimit = MaxExtentRepairLimit )
var ( ErrIncorrectStoreType = errors.New("Incorrect store type") ErrNoSpaceToCreatePartition = errors.New("No disk space to create a data partition") ErrNewSpaceManagerFailed = errors.New("Creater new space manager failed") ErrGetMasterDatanodeInfoFailed = errors.New("Failed to get datanode info from master") LocalIP string MasterClient = masterSDK.NewMasterClient(nil, false) )
var (
AutoRepairStatus = true
)
Functions ¶
func DeleteLimiterWait ¶
func DeleteLimiterWait()
func GetIoMetricLabels ¶
func GetIoMetricLabels(partition *DataPartition, tp string) map[string]string
func MarshalRandWriteRaftLog ¶ added in v1.4.0
func NewPacketToBroadcastMinAppliedID ¶
NewPacketToBroadcastMinAppliedID returns a new packet to broadcast the min applied ID.
func NewPacketToGetAppliedID ¶
NewPacketToGetAppliedID returns a new packet to get the applied ID.
func NewPacketToGetMaxExtentIDAndPartitionSIze ¶ added in v1.4.0
NewPacketToGetPartitionSize returns a new packet to get the partition size.
func NewPacketToGetPartitionSize ¶
NewPacketToGetPartitionSize returns a new packet to get the partition size.
func UnmarshalOldVersionRaftLog ¶ added in v1.4.0
func UnmarshalOldVersionRandWriteOpItem ¶ added in v1.4.0
func UnmarshalRandWriteRaftLog ¶ added in v1.4.0
RandomWriteSubmit submits the proposal to raft.
Types ¶
type DataNode ¶
type DataNode struct {
// contains filtered or unexported fields
}
DataNode defines the structure of a data node.
func (*DataNode) OperatePacket ¶
type DataNodeInfo ¶ added in v1.4.0
type DataNodeMetrics ¶
type DataPartition ¶
type DataPartition struct { DataPartitionCreateType int // contains filtered or unexported fields }
func CreateDataPartition ¶
func CreateDataPartition(dpCfg *dataPartitionCfg, disk *Disk, request *proto.CreateDataPartitionRequest) (dp *DataPartition, err error)
func LoadDataPartition ¶
func LoadDataPartition(partitionDir string, disk *Disk) (dp *DataPartition, err error)
LoadDataPartition loads and returns a partition instance based on the specified directory. It reads the partition metadata file stored under the specified directory and creates the partition instance.
func (*DataPartition) Apply ¶
func (dp *DataPartition) Apply(command []byte, index uint64) (resp interface{}, err error)
Apply puts the data onto the disk.
func (*DataPartition) ApplyMemberChange ¶
func (dp *DataPartition) ApplyMemberChange(confChange *raftproto.ConfChange, index uint64) (resp interface{}, err error)
ApplyMemberChange supports adding new raft member or deleting an existing raft member. It does not support updating an existing member at this point.
func (*DataPartition) ApplyRandomWrite ¶
func (dp *DataPartition) ApplyRandomWrite(command []byte, raftApplyID uint64) (resp interface{}, err error)
ApplyRandomWrite random write apply
func (*DataPartition) ApplySnapshot ¶
func (dp *DataPartition) ApplySnapshot(peers []raftproto.Peer, iterator raftproto.SnapIterator) (err error)
ApplySnapshot asks the raft leader for the snapshot data to recover the contents on the local disk.
func (*DataPartition) Available ¶
func (dp *DataPartition) Available() int
Available returns the available space.
func (*DataPartition) CanRemoveRaftMember ¶ added in v1.4.0
func (dp *DataPartition) CanRemoveRaftMember(peer proto.Peer) error
func (*DataPartition) ChangeRaftMember ¶
func (dp *DataPartition) ChangeRaftMember(changeType raftProto.ConfChangeType, peer raftProto.Peer, context []byte) (resp interface{}, err error)
ChangeRaftMember is a wrapper function of changing the raft member.
func (*DataPartition) CheckLeader ¶
CheckLeader checks if itself is the leader during read
func (*DataPartition) Del ¶
func (dp *DataPartition) Del(key interface{}) (interface{}, error)
Del deletes the raft log based on the given key. It is not needed for replicating data partition.
func (*DataPartition) DoExtentStoreRepair ¶
func (dp *DataPartition) DoExtentStoreRepair(repairTask *DataPartitionRepairTask)
DoExtentStoreRepair performs the repairs of the extent store. 1. when the extent size is smaller than the max size on the record, start to repair the missing part. 2. if the extent does not even exist, create the extent first, and then repair.
func (*DataPartition) DoRepair ¶
func (dp *DataPartition) DoRepair(repairTasks []*DataPartitionRepairTask)
DoRepair asks the leader to perform the repair tasks.
func (*DataPartition) ExtentStore ¶
func (dp *DataPartition) ExtentStore() *storage.ExtentStore
func (*DataPartition) ForceLoadHeader ¶
func (dp *DataPartition) ForceLoadHeader()
func (*DataPartition) ForceSetDataPartitionToFininshLoad ¶
func (dp *DataPartition) ForceSetDataPartitionToFininshLoad()
func (*DataPartition) ForceSetDataPartitionToLoadding ¶
func (dp *DataPartition) ForceSetDataPartitionToLoadding()
func (*DataPartition) ForceSetRaftRunning ¶
func (dp *DataPartition) ForceSetRaftRunning()
func (*DataPartition) Get ¶
func (dp *DataPartition) Get(key interface{}) (interface{}, error)
Get returns the raft log based on the given key. It is not needed for replicating data partition.
func (*DataPartition) GetAppliedID ¶
func (dp *DataPartition) GetAppliedID() (id uint64)
func (*DataPartition) GetExtentCount ¶
func (dp *DataPartition) GetExtentCount() int
func (*DataPartition) HandleFatalEvent ¶
func (dp *DataPartition) HandleFatalEvent(err *raft.FatalError)
HandleFatalEvent notifies the application when panic happens.
func (*DataPartition) HandleLeaderChange ¶
func (dp *DataPartition) HandleLeaderChange(leader uint64)
HandleLeaderChange notifies the application when the raft leader has changed.
func (*DataPartition) IsEquareCreateDataPartitionRequst ¶
func (dp *DataPartition) IsEquareCreateDataPartitionRequst(request *proto.CreateDataPartitionRequest) (err error)
func (*DataPartition) IsExsitReplica ¶ added in v1.4.0
func (dp *DataPartition) IsExsitReplica(addr string) bool
func (*DataPartition) IsRaftLeader ¶
func (dp *DataPartition) IsRaftLeader() (addr string, ok bool)
IsRaftLeader tells if the given address belongs to the raft leader.
func (*DataPartition) IsRejectWrite ¶ added in v1.4.0
func (dp *DataPartition) IsRejectWrite() bool
func (*DataPartition) LaunchRepair ¶
func (dp *DataPartition) LaunchRepair(extentType uint8)
LaunchRepair launches the repair of extents.
func (*DataPartition) Load ¶
func (dp *DataPartition) Load() (response *proto.LoadDataPartitionResponse)
func (*DataPartition) LoadAppliedID ¶
func (dp *DataPartition) LoadAppliedID() (err error)
LoadAppliedID loads the applied IDs to the memory.
func (*DataPartition) NotifyExtentRepair ¶
func (dp *DataPartition) NotifyExtentRepair(members []*DataPartitionRepairTask) (err error)
NotifyExtentRepair notifies the followers to repair.
func (*DataPartition) Path ¶
func (dp *DataPartition) Path() string
func (*DataPartition) PersistMetadata ¶
func (dp *DataPartition) PersistMetadata() (err error)
PersistMetadata persists the file metadata on the disk.
func (*DataPartition) Put ¶
func (dp *DataPartition) Put(key interface{}, val interface{}) (resp interface{}, err error)
Put submits the raft log to the raft store.
func (*DataPartition) RandomWriteSubmit ¶
func (dp *DataPartition) RandomWriteSubmit(pkg *repl.Packet) (err error)
RandomWriteSubmit submits the proposal to raft.
func (*DataPartition) ReloadSnapshot ¶
func (dp *DataPartition) ReloadSnapshot()
func (*DataPartition) Replicas ¶
func (dp *DataPartition) Replicas() []string
func (*DataPartition) SetMinAppliedID ¶
func (dp *DataPartition) SetMinAppliedID(id uint64)
func (*DataPartition) SnapShot ¶
func (dp *DataPartition) SnapShot() (files []*proto.File)
Snapshot returns the snapshot of the data partition.
func (*DataPartition) Snapshot ¶
func (dp *DataPartition) Snapshot() (raftproto.Snapshot, error)
Snapshot persists the in-memory data (as a snapshot) to the disk. Note that the data in each data partition has already been saved on the disk. Therefore there is no need to take the snapshot in this case.
func (*DataPartition) StartRaft ¶
func (dp *DataPartition) StartRaft() (err error)
StartRaft start raft instance when data partition start or restore.
func (*DataPartition) StartRaftAfterRepair ¶
func (dp *DataPartition) StartRaftAfterRepair()
StartRaftAfterRepair starts the raft after repairing a partition. It can only happens after all the extent files are repaired by the leader. When the repair is finished, the local dp.partitionSize is same as the leader's dp.partitionSize. The repair task can be done in statusUpdateScheduler->LaunchRepair.
func (*DataPartition) StartRaftLoggingSchedule ¶
func (dp *DataPartition) StartRaftLoggingSchedule()
StartRaftLoggingSchedule starts the task schedule as follows: 1. write the raft applied id into disk. 2. collect the applied ids from raft members. 3. based on the minimum applied id to cutoff and delete the saved raft log in order to free the disk space.
func (*DataPartition) Status ¶
func (dp *DataPartition) Status() int
Status returns the partition status.
func (*DataPartition) Stop ¶
func (dp *DataPartition) Stop()
Stop close the store and the raft store.
func (*DataPartition) String ¶
func (dp *DataPartition) String() (m string)
String returns the string format of the data partition information.
type DataPartitionMetadata ¶
type DataPartitionMetadata struct { VolumeID string PartitionID uint64 PartitionSize int PartitionType int CreateTime string Peers []proto.Peer Hosts []string DataPartitionCreateType int LastTruncateID uint64 }
func (*DataPartitionMetadata) Validate ¶
func (md *DataPartitionMetadata) Validate() (err error)
type DataPartitionRepairTask ¶
type DataPartitionRepairTask struct { TaskType uint8 ExtentsToBeCreated []*storage.ExtentInfo ExtentsToBeRepaired []*storage.ExtentInfo LeaderTinyDeleteRecordFileSize int64 LeaderAddr string // contains filtered or unexported fields }
DataPartitionRepairTask defines the reapir task for the data partition.
func NewDataPartitionRepairTask ¶
func NewDataPartitionRepairTask(extentFiles []*storage.ExtentInfo, tinyDeleteRecordFileSize int64, source, leaderAddr string) (task *DataPartitionRepairTask)
type Disk ¶
type Disk struct { sync.RWMutex Path string ReadErrCnt uint64 // number of read errors WriteErrCnt uint64 // number of write errors Total uint64 Used uint64 Available uint64 Unallocated uint64 Allocated uint64 MaxErrCnt int // maximum number of errors Status int // disk status such as READONLY ReservedSpace uint64 RejectWrite bool // contains filtered or unexported fields }
Disk represents the structure of the disk
func NewDisk ¶
func NewDisk(path string, reservedSpace uint64, maxErrCnt int, space *SpaceManager) (d *Disk)
func (*Disk) AttachDataPartition ¶
func (d *Disk) AttachDataPartition(dp *DataPartition)
AttachDataPartition adds a data partition to the partition map.
func (*Disk) DataPartitionList ¶
DataPartitionList returns a list of the data partitions
func (*Disk) DetachDataPartition ¶
func (d *Disk) DetachDataPartition(dp *DataPartition)
DetachDataPartition removes a data partition from the partition map.
func (*Disk) ForceExitRaftStore ¶ added in v1.1.1
func (d *Disk) ForceExitRaftStore()
func (*Disk) GetDataPartition ¶
func (d *Disk) GetDataPartition(partitionID uint64) (partition *DataPartition)
GetDataPartition returns the data partition based on the given partition ID.
func (*Disk) PartitionCount ¶
PartitionCount returns the number of partitions in the partition map.
func (*Disk) RestorePartition ¶
func (d *Disk) RestorePartition(visitor PartitionVisitor)
RestorePartition reads the files stored on the local disk and restores the data partitions.
type ItemIterator ¶
type ItemIterator struct {
// contains filtered or unexported fields
}
func NewItemIterator ¶
func NewItemIterator(applyID uint64) *ItemIterator
NewItemIterator creates a new item iterator.
func (*ItemIterator) ApplyIndex ¶
func (si *ItemIterator) ApplyIndex() uint64
ApplyIndex returns the appliedID
func (*ItemIterator) Next ¶
func (si *ItemIterator) Next() (data []byte, err error)
Next returns the next item in the iterator.
type PartitionVisitor ¶
type PartitionVisitor func(dp *DataPartition)
type RaftCmdItem ¶
type SimpleVolView ¶
type SimpleVolView struct {
// contains filtered or unexported fields
}
type SpaceManager ¶
type SpaceManager struct {
// contains filtered or unexported fields
}
SpaceManager manages the disk space.
func NewSpaceManager ¶
func NewSpaceManager(dataNode *DataNode) *SpaceManager
NewSpaceManager creates a new space manager.
func (*SpaceManager) AttachPartition ¶ added in v1.4.0
func (manager *SpaceManager) AttachPartition(dp *DataPartition)
func (*SpaceManager) CreatePartition ¶
func (manager *SpaceManager) CreatePartition(request *proto.CreateDataPartitionRequest) (dp *DataPartition, err error)
func (*SpaceManager) DeletePartition ¶
func (manager *SpaceManager) DeletePartition(dpID uint64)
DeletePartition deletes a partition based on the partition id.
func (*SpaceManager) DetachDataPartition ¶ added in v1.4.0
func (manager *SpaceManager) DetachDataPartition(partitionID uint64)
DetachDataPartition removes a data partition from the partition map.
func (*SpaceManager) GetClusterID ¶
func (manager *SpaceManager) GetClusterID() (clusterID string)
func (*SpaceManager) GetDisk ¶
func (manager *SpaceManager) GetDisk(path string) (d *Disk, err error)
func (*SpaceManager) GetDisks ¶
func (manager *SpaceManager) GetDisks() (disks []*Disk)
func (*SpaceManager) GetNodeID ¶
func (manager *SpaceManager) GetNodeID() (nodeID uint64)
func (*SpaceManager) GetRaftStore ¶
func (manager *SpaceManager) GetRaftStore() (raftStore raftstore.RaftStore)
func (*SpaceManager) LoadDisk ¶
func (manager *SpaceManager) LoadDisk(path string, reservedSpace uint64, maxErrCnt int) (err error)
func (*SpaceManager) Partition ¶
func (manager *SpaceManager) Partition(partitionID uint64) (dp *DataPartition)
func (*SpaceManager) RangePartitions ¶
func (manager *SpaceManager) RangePartitions(f func(partition *DataPartition) bool)
func (*SpaceManager) SetClusterID ¶
func (manager *SpaceManager) SetClusterID(clusterID string)
func (*SpaceManager) SetNodeID ¶
func (manager *SpaceManager) SetNodeID(nodeID uint64)
func (*SpaceManager) SetRaftStore ¶
func (manager *SpaceManager) SetRaftStore(raftStore raftstore.RaftStore)
func (*SpaceManager) Stats ¶
func (manager *SpaceManager) Stats() *Stats
func (*SpaceManager) Stop ¶
func (manager *SpaceManager) Stop()
type Stats ¶
type Stats struct { Zone string ConnectionCnt int64 ClusterID string TCPAddr string Start time.Time Total uint64 Used uint64 Available uint64 // available space TotalPartitionSize uint64 // dataPartitionCnt * dataPartitionSize RemainingCapacityToCreatePartition uint64 CreatedPartitionCnt uint64 // the maximum capacity among all the nodes that can be used to create partition MaxCapacityToCreatePartition uint64 sync.Mutex // contains filtered or unexported fields }
Stats defines various metrics that will be collected during the execution.
func (*Stats) GetConnectionCount ¶
GetConnectionCount gets the connection count.
func (*Stats) RemoveConnection ¶
func (s *Stats) RemoveConnection()
RemoveConnection removes a connection.