Documentation
¶
Index ¶
- func AdvanceTicks(n etcdraft.Node, electionTicks int)
- func MustGetAppliedIndex(s storage.Storage) uint64
- type ApplyEntry
- type Node
- func (r *Node) Apply() chan ApplyEntry
- func (r *Node) DoneC() chan struct{}
- func (r *Node) GetAppliedIndex() uint64
- func (r *Node) GetCommittedIndex() uint64
- func (r *Node) GetTerm() uint64
- func (r *Node) MsgSnapC() chan raftpb.Message
- func (r *Node) NewPeerListener(peerURLs types.URLs) (err error)
- func (r *Node) PeerHandlerServe()
- func (r *Node) Run(rh *ReadyHandler)
- func (r *Node) SetAppliedIndex(v uint64)
- func (r *Node) SetCommittedIndex(v uint64)
- func (r *Node) SetPeerHandler()
- func (r *Node) SetTerm(v uint64)
- func (r *Node) Stop()
- func (r *Node) StopPeerListener() error
- type NodeConfig
- type ReadyHandler
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func AdvanceTicks ¶
AdvanceTicks advances ticks of Raft node. This can be used for fast-forwarding election ticks in multi data-center deployments, thus speeding up election process.
func MustGetAppliedIndex ¶
MustGetAppliedIndex get the applied raft index, if occur error, fatal
Types ¶
type ApplyEntry ¶
type ApplyEntry struct {
Entries []raftpb.Entry
Snapshot raftpb.Snapshot
// Notifyc synchronizes etcd server applies with the raft node
Notifyc chan struct{}
}
ApplyEntry contains entries, Snapshot to be applied. Once an ApplyEntry is consumed, the entries will be persisted to to raft storage concurrently; the application must read raftDone before assuming the raft messages are stable.
type Node ¶
type Node struct {
NodeConfig
// contains filtered or unexported fields
}
Node represents a raft node
func (*Node) GetAppliedIndex ¶
GetAppliedIndex get the applied raft index
func (*Node) GetCommittedIndex ¶
GetCommittedIndex get the committed raft index
func (*Node) NewPeerListener ¶
NewPeerListener create listener of peer
func (*Node) Run ¶
func (r *Node) Run(rh *ReadyHandler)
Run must keep run as fast as possible If execution is too slow, ticker will suffer, then tickHeartbeat(tickElection) will be influenced
func (*Node) SetAppliedIndex ¶
SetAppliedIndex set the applied raft index, not used
func (*Node) SetCommittedIndex ¶
SetCommittedIndex set the committed raft index
func (*Node) SetPeerHandler ¶
func (r *Node) SetPeerHandler()
SetPeerHandler set the handler of raft transport
func (*Node) StopPeerListener ¶
StopPeerListener stop the listener of peer
type NodeConfig ¶
type NodeConfig struct {
// to check if msg receiver is removed from cluster
IsIDRemoved func(id uint64) bool
etcdraft.Node
//raftStorage *raft.MemoryStorage
//storage Storage
//store raft log
Storage storage.Storage
//MetadataStorage MetadataStorage
Heartbeat time.Duration // for logging
// Transport specifies the Transport to send and receive msgs to members.
// Sending messages MUST NOT block. It is okay to drop messages, since
// clients should timeout and reissue their messages.
// If Transport is nil, server will panic.
Transport rafthttp.Transporter
PeerListener []*peerListener
}
NodeConfig is the config of raft node
type ReadyHandler ¶
type ReadyHandler struct {
GetLead func() (lead uint64)
UpdateLead func(lead uint64)
UpdateLeadership func(leadChange bool)
UpdateCommittedIndex func(uint64)
}
ReadyHandler contains a set of EtcdServer operations to be called by Node, and helps decouple state machine logic from Raft algorithms. TODO: add a state machine interface to ApplyEntry the commit entries and do Snapshot/recover
Source Files
Directories