v0.0.0-...-d88c8b5 Latest Latest

This package is not in the latest version of its module.

Go to latest
Published: Feb 9, 2021 License: Apache-2.0 Imports: 17 Imported by: 0



Package raft sends and receives messages in the Protocol Buffer format defined in the raftpb package.

Raft is a protocol with which a cluster of nodes can maintain a replicated state machine. The state machine is kept in sync through the use of a replicated log. For more details on Raft, see "In Search of an Understandable Consensus Algorithm" (https://raft.github.io/raft.pdf) by Diego Ongaro and John Ousterhout.

A simple example application, _raftexample_, is also available to help illustrate how to use this package in practice: https://github.com/etcd-io/etcd/tree/master/contrib/raftexample


The primary object in raft is a Node. You either start a Node from scratch using raft.StartNode or start a Node from some initial state using raft.RestartNode.

To start a node from scratch:

storage := raft.NewMemoryStorage()
c := &Config{
  ID:              0x01,
  ElectionTick:    10,
  HeartbeatTick:   1,
  Storage:         storage,
  MaxSizePerMsg:   4096,
  MaxInflightMsgs: 256,
n := raft.StartNode(c, []raft.Peer{{ID: 0x02}, {ID: 0x03}})

To restart a node from previous state:

storage := raft.NewMemoryStorage()

// recover the in-memory storage from persistent
// snapshot, state and entries.

c := &Config{
  ID:              0x01,
  ElectionTick:    10,
  HeartbeatTick:   1,
  Storage:         storage,
  MaxSizePerMsg:   4096,
  MaxInflightMsgs: 256,

// restart raft without peer information.
// peer information is already included in the storage.
n := raft.RestartNode(c)

Now that you are holding onto a Node you have a few responsibilities:

First, you must read from the Node.Ready() channel and process the updates it contains. These steps may be performed in parallel, except as noted in step 2.

1. Write HardState, Entries, and Snapshot to persistent storage if they are not empty. Note that when writing an Entry with Index i, any previously-persisted entries with Index >= i must be discarded.

2. Send all Messages to the nodes named in the To field. It is important that no messages be sent until the latest HardState has been persisted to disk, and all Entries written by any previous Ready batch (Messages may be sent while entries from the same batch are being persisted). To reduce the I/O latency, an optimization can be applied to make leader write to disk in parallel with its followers (as explained at section 10.2.1 in Raft thesis). If any Message has type MsgSnap, call Node.ReportSnapshot() after it has been sent (these messages may be large).

Note: Marshalling messages is not thread-safe; it is important that you make sure that no new entries are persisted while marshalling. The easiest way to achieve this is to serialize the messages directly inside your main raft loop.

3. Apply Snapshot (if any) and CommittedEntries to the state machine. If any committed Entry has Type EntryConfChange, call Node.ApplyConfChange() to apply it to the node. The configuration change may be cancelled at this point by setting the NodeID field to zero before calling ApplyConfChange (but ApplyConfChange must be called one way or the other, and the decision to cancel must be based solely on the state machine and not external information such as the observed health of the node).

4. Call Node.Advance() to signal readiness for the next batch of updates. This may be done at any time after step 1, although all updates must be processed in the order they were returned by Ready.

Second, all persisted log entries must be made available via an implementation of the Storage interface. The provided MemoryStorage type can be used for this (if you repopulate its state upon a restart), or you can supply your own disk-backed implementation.

Third, when you receive a message from another node, pass it to Node.Step:

func recvRaftRPC(ctx context.Context, m raftpb.Message) {
	n.Step(ctx, m)

Finally, you need to call Node.Tick() at regular intervals (probably via a time.Ticker). Raft has two important timeouts: heartbeat and the election timeout. However, internally to the raft package time is represented by an abstract "tick".

The total state machine handling loop will look something like this:

for {
  select {
  case <-s.Ticker:
  case rd := <-s.Node.Ready():
    saveToStorage(rd.State, rd.Entries, rd.Snapshot)
    if !raft.IsEmptySnap(rd.Snapshot) {
    for _, entry := range rd.CommittedEntries {
      if entry.Type == raftpb.EntryConfChange {
        var cc raftpb.ConfChange
  case <-s.done:

To propose changes to the state machine from your node take your application data, serialize it into a byte slice and call:

n.Propose(ctx, data)

If the proposal is committed, data will appear in committed entries with type raftpb.EntryNormal. There is no guarantee that a proposed command will be committed; you may have to re-propose after a timeout.

To add or remove a node in a cluster, build ConfChange struct 'cc' and call:

n.ProposeConfChange(ctx, cc)

After config change is committed, some committed entry with type raftpb.EntryConfChange will be returned. You must apply it to node through:

var cc raftpb.ConfChange

Note: An ID represents a unique node in a cluster for all time. A given ID MUST be used only once even if the old node has been removed. This means that for example IP addresses make poor node IDs since they may be reused. Node IDs must be non-zero.

Implementation notes

This implementation is up to date with the final Raft thesis (https://github.com/ongardie/dissertation/blob/master/stanford.pdf), although our implementation of the membership change protocol differs somewhat from that described in chapter 4. The key invariant that membership changes happen one node at a time is preserved, but in our implementation the membership change takes effect when its entry is applied, not when it is added to the log (so the entry is committed under the old membership instead of the new). This is equivalent in terms of safety, since the old and new configurations are guaranteed to overlap.

To ensure that we do not attempt to commit two membership changes at once by matching log positions (which would be unsafe since they should have different quorum requirements), we simply disallow any proposed membership change while any uncommitted change appears in the leader's log.

This approach introduces a problem when you try to remove a member from a two-member cluster: If one of the members dies before the other one receives the commit of the confchange entry, then the member cannot be removed any more since the cluster cannot make progress. For this reason it is highly recommended to use three or more nodes in every cluster.


Package raft sends and receives message in Protocol Buffer format (defined in raftpb package). Each state (follower, candidate, leader) implements its own 'step' method ('stepFollower', 'stepCandidate', 'stepLeader') when advancing with the given raftpb.Message. Each step is determined by its raftpb.MessageType. Note that every step is checked by one common method 'Step' that safety-checks the terms of node and incoming message to prevent stale log entries:

'MsgHup' is used for election. If a node is a follower or candidate, the
'tick' function in 'raft' struct is set as 'tickElection'. If a follower or
candidate has not received any heartbeat before the election timeout, it
passes 'MsgHup' to its Step method and becomes (or remains) a candidate to
start a new election.

'MsgBeat' is an internal type that signals the leader to send a heartbeat of
the 'MsgHeartbeat' type. If a node is a leader, the 'tick' function in
the 'raft' struct is set as 'tickHeartbeat', and triggers the leader to
send periodic 'MsgHeartbeat' messages to its followers.

'MsgProp' proposes to append data to its log entries. This is a special
type to redirect proposals to leader. Therefore, send method overwrites
raftpb.Message's term with its HardState's term to avoid attaching its
local term to 'MsgProp'. When 'MsgProp' is passed to the leader's 'Step'
method, the leader first calls the 'appendEntry' method to append entries
to its log, and then calls 'bcastAppend' method to send those entries to
its peers. When passed to candidate, 'MsgProp' is dropped. When passed to
follower, 'MsgProp' is stored in follower's mailbox(msgs) by the send
method. It is stored with sender's ID and later forwarded to leader by
rafthttp package.

'MsgApp' contains log entries to replicate. A leader calls bcastAppend,
which calls sendAppend, which sends soon-to-be-replicated logs in 'MsgApp'
type. When 'MsgApp' is passed to candidate's Step method, candidate reverts
back to follower, because it indicates that there is a valid leader sending
'MsgApp' messages. Candidate and follower respond to this message in
'MsgAppResp' type.

'MsgAppResp' is response to log replication request('MsgApp'). When
'MsgApp' is passed to candidate or follower's Step method, it responds by
calling 'handleAppendEntries' method, which sends 'MsgAppResp' to raft

'MsgVote' requests votes for election. When a node is a follower or
candidate and 'MsgHup' is passed to its Step method, then the node calls
'campaign' method to campaign itself to become a leader. Once 'campaign'
method is called, the node becomes candidate and sends 'MsgVote' to peers
in cluster to request votes. When passed to leader or candidate's Step
method and the message's Term is lower than leader's or candidate's,
'MsgVote' will be rejected ('MsgVoteResp' is returned with Reject true).
If leader or candidate receives 'MsgVote' with higher term, it will revert
back to follower. When 'MsgVote' is passed to follower, it votes for the
sender only when sender's last term is greater than MsgVote's term or
sender's last term is equal to MsgVote's term but sender's last committed
index is greater than or equal to follower's.

'MsgVoteResp' contains responses from voting request. When 'MsgVoteResp' is
passed to candidate, the candidate calculates how many votes it has won. If
it's more than majority (quorum), it becomes leader and calls 'bcastAppend'.
If candidate receives majority of votes of denials, it reverts back to

'MsgPreVote' and 'MsgPreVoteResp' are used in an optional two-phase election
protocol. When Config.PreVote is true, a pre-election is carried out first
(using the same rules as a regular election), and no node increases its term
number unless the pre-election indicates that the campaigning node would win.
This minimizes disruption when a partitioned node rejoins the cluster.

'MsgSnap' requests to install a snapshot message. When a node has just
become a leader or the leader receives 'MsgProp' message, it calls
'bcastAppend' method, which then calls 'sendAppend' method to each
follower. In 'sendAppend', if a leader fails to get term or entries,
the leader requests snapshot by sending 'MsgSnap' type message.

'MsgSnapStatus' tells the result of snapshot install message. When a
follower rejected 'MsgSnap', it indicates the snapshot request with
'MsgSnap' had failed from network issues which causes the network layer
to fail to send out snapshots to its followers. Then leader considers
follower's progress as probe. When 'MsgSnap' were not rejected, it
indicates that the snapshot succeeded and the leader sets follower's
progress to probe and resumes its log replication.

'MsgHeartbeat' sends heartbeat from leader. When 'MsgHeartbeat' is passed
to candidate and message's term is higher than candidate's, the candidate
reverts back to follower and updates its committed index from the one in
this heartbeat. And it sends the message to its mailbox. When
'MsgHeartbeat' is passed to follower's Step method and message's term is
higher than follower's, the follower updates its leaderID with the ID
from the message.

'MsgHeartbeatResp' is a response to 'MsgHeartbeat'. When 'MsgHeartbeatResp'
is passed to leader's Step method, the leader knows which follower
responded. And only when the leader's last committed index is greater than
follower's Match index, the leader runs 'sendAppend` method.

'MsgUnreachable' tells that request(message) wasn't delivered. When
'MsgUnreachable' is passed to leader's Step method, the leader discovers
that the follower that sent this 'MsgUnreachable' is not reachable, often
indicating 'MsgApp' is lost. When follower's progress state is replicate,
the leader sets it back to probe.



View Source
const None uint64 = 0

None is a placeholder node ID used when there is no leader.


View Source
var ErrCompacted = errors.New("requested index is unavailable due to compaction")

ErrCompacted is returned by Storage.Entries/Compact when a requested index is unavailable because it predates the last snapshot.

View Source
var ErrProposalDropped = errors.New("raft proposal dropped")

ErrProposalDropped is returned when the proposal is ignored by some cases, so that the proposer can be notified and fail fast.

View Source
var ErrSnapOutOfDate = errors.New("requested index is older than the existing snapshot")

ErrSnapOutOfDate is returned by Storage.CreateSnapshot when a requested index is older than the existing snapshot.

View Source
var ErrSnapshotTemporarilyUnavailable = errors.New("snapshot is temporarily unavailable")

ErrSnapshotTemporarilyUnavailable is returned by the Storage interface when the required snapshot is temporarily unavailable.

View Source
var ErrStepLocalMsg = errors.New("raft: cannot step raft local message")

ErrStepLocalMsg is returned when try to step a local raft message

View Source
var ErrStepPeerNotFound = errors.New("raft: cannot step as peer not found")

ErrStepPeerNotFound is returned when try to step a response message but there is no peer found in raft.prs for that node.

View Source
var (

	// ErrStopped is returned by methods on Nodes that have been stopped.
	ErrStopped = errors.New("raft: stopped")
View Source
var ErrUnavailable = errors.New("requested entry at index is unavailable")

ErrUnavailable is returned by Storage interface when the requested log entries are unavailable.


func DescribeConfState

func DescribeConfState(state pb.ConfState) string

func DescribeEntries

func DescribeEntries(ents []pb.Entry, f EntryFormatter) string

DescribeEntries calls DescribeEntry for each Entry, adding a newline to each.

func DescribeEntry

func DescribeEntry(e pb.Entry, f EntryFormatter) string

DescribeEntry returns a concise human-readable description of an Entry for debugging.

func DescribeHardState

func DescribeHardState(hs pb.HardState) string

func DescribeMessage

func DescribeMessage(m pb.Message, f EntryFormatter) string

DescribeMessage returns a concise human-readable description of a Message for debugging.

func DescribeReady

func DescribeReady(rd Ready, f EntryFormatter) string

func DescribeSnapshot

func DescribeSnapshot(snap pb.Snapshot) string

func DescribeSoftState

func DescribeSoftState(ss SoftState) string

func IsEmptyHardState

func IsEmptyHardState(st pb.HardState) bool

IsEmptyHardState returns true if the given HardState is empty.

func IsEmptySnap

func IsEmptySnap(sp pb.Snapshot) bool

IsEmptySnap returns true if the given Snapshot is empty.

func IsLocalMsg

func IsLocalMsg(msgt pb.MessageType) bool

func IsResponseMsg

func IsResponseMsg(msgt pb.MessageType) bool

func MustSync

func MustSync(st, prevst pb.HardState, entsnum int) bool

MustSync returns true if the hard state and count of Raft entries indicate that a synchronous write to persistent storage is required.

func PayloadSize

func PayloadSize(e pb.Entry) int

PayloadSize is the size of the payload of this Entry. Notably, it does not depend on its Index or Term.

func SetLogger

func SetLogger(l Logger)


type BasicStatus

type BasicStatus struct {
	ID uint64


	Applied uint64

	LeadTransferee uint64

BasicStatus contains basic information about the Raft peer. It does not allocate.

type CampaignType

type CampaignType string

CampaignType represents the type of campaigning the reason we use the type of string instead of uint64 is because it's simpler to compare and fill in raft entries

type Config

type Config struct {
	// ID is the identity of the local raft. ID cannot be 0.
	ID uint64

	// ElectionTick is the number of Node.Tick invocations that must pass between
	// elections. That is, if a follower does not receive any message from the
	// leader of current term before ElectionTick has elapsed, it will become
	// candidate and start an election. ElectionTick must be greater than
	// HeartbeatTick. We suggest ElectionTick = 10 * HeartbeatTick to avoid
	// unnecessary leader switching.
	ElectionTick int
	// HeartbeatTick is the number of Node.Tick invocations that must pass between
	// heartbeats. That is, a leader sends heartbeat messages to maintain its
	// leadership every HeartbeatTick ticks.
	HeartbeatTick int

	// Storage is the storage for raft. raft generates entries and states to be
	// stored in storage. raft reads the persisted entries and states out of
	// Storage when it needs. raft reads out the previous state and configuration
	// out of storage when restarting.
	Storage Storage
	// Applied is the last applied index. It should only be set when restarting
	// raft. raft will not return entries to the application smaller or equal to
	// Applied. If Applied is unset when restarting, raft might return previous
	// applied entries. This is a very application dependent configuration.
	Applied uint64

	// MaxSizePerMsg limits the max byte size of each append message. Smaller
	// value lowers the raft recovery cost(initial probing and message lost
	// during normal operation). On the other side, it might affect the
	// throughput during normal replication. Note: math.MaxUint64 for unlimited,
	// 0 for at most one entry per message.
	MaxSizePerMsg uint64
	// MaxCommittedSizePerReady limits the size of the committed entries which
	// can be applied.
	MaxCommittedSizePerReady uint64
	// MaxUncommittedEntriesSize limits the aggregate byte size of the
	// uncommitted entries that may be appended to a leader's log. Once this
	// limit is exceeded, proposals will begin to return ErrProposalDropped
	// errors. Note: 0 for no limit.
	MaxUncommittedEntriesSize uint64
	// MaxInflightMsgs limits the max number of in-flight append messages during
	// optimistic replication phase. The application transportation layer usually
	// has its own sending buffer over TCP/UDP. Setting MaxInflightMsgs to avoid
	// overflowing that sending buffer. TODO (xiangli): feedback to application to
	// limit the proposal rate?
	MaxInflightMsgs int

	// CheckQuorum specifies if the leader should check quorum activity. Leader
	// steps down when quorum is not active for an electionTimeout.
	CheckQuorum bool

	// PreVote enables the Pre-Vote algorithm described in raft thesis section
	// 9.6. This prevents disruption when a node that has been partitioned away
	// rejoins the cluster.
	PreVote bool

	// ReadOnlyOption specifies how the read only request is processed.
	// ReadOnlySafe guarantees the linearizability of the read only request by
	// communicating with the quorum. It is the default and suggested option.
	// ReadOnlyLeaseBased ensures linearizability of the read only request by
	// relying on the leader lease. It can be affected by clock drift.
	// If the clock drift is unbounded, leader might keep the lease longer than it
	// should (clock can move backward/pause without any bound). ReadIndex is not safe
	// in that case.
	// CheckQuorum MUST be enabled if ReadOnlyOption is ReadOnlyLeaseBased.
	ReadOnlyOption ReadOnlyOption

	// Logger is the logger used for raft log. For multinode which can host
	// multiple raft group, each raft group can have its own logger
	Logger Logger

	// DisableProposalForwarding set to true means that followers will drop
	// proposals, rather than forwarding them to the leader. One use case for
	// this feature would be in a situation where the Raft leader is used to
	// compute the data of a proposal, for example, adding a timestamp from a
	// hybrid logical clock to data in a monotonically increasing way. Forwarding
	// should be disabled to prevent a follower with an inaccurate hybrid
	// logical clock from assigning the timestamp and then forwarding the data
	// to the leader.
	DisableProposalForwarding bool
	// contains filtered or unexported fields

Config contains the parameters to start a raft.

type DefaultLogger

type DefaultLogger struct {
	// contains filtered or unexported fields

DefaultLogger is a default implementation of the Logger interface.

func (*DefaultLogger) Debug

func (l *DefaultLogger) Debug(v ...interface{})

func (*DefaultLogger) Debugf

func (l *DefaultLogger) Debugf(format string, v ...interface{})

func (*DefaultLogger) EnableDebug

func (l *DefaultLogger) EnableDebug()

func (*DefaultLogger) EnableTimestamps

func (l *DefaultLogger) EnableTimestamps()

func (*DefaultLogger) Error

func (l *DefaultLogger) Error(v ...interface{})

func (*DefaultLogger) Errorf

func (l *DefaultLogger) Errorf(format string, v ...interface{})

func (*DefaultLogger) Fatal

func (l *DefaultLogger) Fatal(v ...interface{})

func (*DefaultLogger) Fatalf

func (l *DefaultLogger) Fatalf(format string, v ...interface{})

func (*DefaultLogger) Info

func (l *DefaultLogger) Info(v ...interface{})

func (*DefaultLogger) Infof

func (l *DefaultLogger) Infof(format string, v ...interface{})

func (*DefaultLogger) Panic

func (l *DefaultLogger) Panic(v ...interface{})

func (*DefaultLogger) Panicf

func (l *DefaultLogger) Panicf(format string, v ...interface{})

func (*DefaultLogger) Warning

func (l *DefaultLogger) Warning(v ...interface{})

func (*DefaultLogger) Warningf

func (l *DefaultLogger) Warningf(format string, v ...interface{})

type EntryFormatter

type EntryFormatter func([]byte) string

EntryFormatter can be implemented by the application to provide human-readable formatting of entry data. Nil is a valid EntryFormatter and will use a default format.

type Logger

type Logger interface {
	Debug(v ...interface{})
	Debugf(format string, v ...interface{})

	Error(v ...interface{})
	Errorf(format string, v ...interface{})

	Info(v ...interface{})
	Infof(format string, v ...interface{})

	Warning(v ...interface{})
	Warningf(format string, v ...interface{})

	Fatal(v ...interface{})
	Fatalf(format string, v ...interface{})

	Panic(v ...interface{})
	Panicf(format string, v ...interface{})

type MemoryStorage

type MemoryStorage struct {
	// Protects access to all fields. Most methods of MemoryStorage are
	// run on the raft goroutine, but Append() is run on an application
	// goroutine.
	// contains filtered or unexported fields

MemoryStorage implements the Storage interface backed by an in-memory array.

func NewMemoryStorage

func NewMemoryStorage() *MemoryStorage

NewMemoryStorage creates an empty MemoryStorage.

func (*MemoryStorage) Append

func (ms *MemoryStorage) Append(entries []pb.Entry) error

Append the new entries to storage. TODO (xiangli): ensure the entries are continuous and entries[0].Index > ms.entries[0].Index

func (*MemoryStorage) ApplySnapshot

func (ms *MemoryStorage) ApplySnapshot(snap pb.Snapshot) error

ApplySnapshot overwrites the contents of this Storage object with those of the given snapshot.

func (*MemoryStorage) Compact

func (ms *MemoryStorage) Compact(compactIndex uint64) error

Compact discards all log entries prior to compactIndex. It is the application's responsibility to not attempt to compact an index greater than raftLog.applied.

func (*MemoryStorage) CreateSnapshot

func (ms *MemoryStorage) CreateSnapshot(i uint64, cs *pb.ConfState, data []byte) (pb.Snapshot, error)

CreateSnapshot makes a snapshot which can be retrieved with Snapshot() and can be used to reconstruct the state at that point. If any configuration changes have been made since the last compaction, the result of the last ApplyConfChange must be passed in.

func (*MemoryStorage) Entries

func (ms *MemoryStorage) Entries(lo, hi, maxSize uint64) ([]pb.Entry, error)

Entries implements the Storage interface.

func (*MemoryStorage) FirstIndex

func (ms *MemoryStorage) FirstIndex() (uint64, error)

FirstIndex implements the Storage interface.

func (*MemoryStorage) InitialState

func (ms *MemoryStorage) InitialState() (pb.HardState, pb.ConfState, error)

InitialState implements the Storage interface.

func (*MemoryStorage) LastIndex

func (ms *MemoryStorage) LastIndex() (uint64, error)

LastIndex implements the Storage interface.

func (*MemoryStorage) SetHardState

func (ms *MemoryStorage) SetHardState(st pb.HardState) error

SetHardState saves the current HardState.

func (*MemoryStorage) Snapshot

func (ms *MemoryStorage) Snapshot() (pb.Snapshot, error)

Snapshot implements the Storage interface.

func (*MemoryStorage) Term

func (ms *MemoryStorage) Term(i uint64) (uint64, error)

Term implements the Storage interface.

type Node

type Node interface {
	// Tick increments the internal logical clock for the Node by a single tick. Election
	// timeouts and heartbeat timeouts are in units of ticks.
	// Campaign causes the Node to transition to candidate state and start campaigning to become leader.
	Campaign(ctx context.Context) error
	// Propose proposes that data be appended to the log. Note that proposals can be lost without
	// notice, therefore it is user's job to ensure proposal retries.
	Propose(ctx context.Context, data []byte) error
	// ProposeConfChange proposes a configuration change. Like any proposal, the
	// configuration change may be dropped with or without an error being
	// returned. In particular, configuration changes are dropped unless the
	// leader has certainty that there is no prior unapplied configuration
	// change in its log.
	// The method accepts either a pb.ConfChange (deprecated) or pb.ConfChangeV2
	// message. The latter allows arbitrary configuration changes via joint
	// consensus, notably including replacing a voter. Passing a ConfChangeV2
	// message is only allowed if all Nodes participating in the cluster run a
	// version of this library aware of the V2 API. See pb.ConfChangeV2 for
	// usage details and semantics.
	ProposeConfChange(ctx context.Context, cc pb.ConfChangeI) error

	// Step advances the state machine using the given message. ctx.Err() will be returned, if any.
	Step(ctx context.Context, msg pb.Message) error

	// Ready returns a channel that returns the current point-in-time state.
	// Users of the Node must call Advance after retrieving the state returned by Ready.
	// NOTE: No committed entries from the next Ready may be applied until all committed entries
	// and snapshots from the previous one have finished.
	Ready() <-chan Ready

	// Advance notifies the Node that the application has saved progress up to the last Ready.
	// It prepares the node to return the next available Ready.
	// The application should generally call Advance after it applies the entries in last Ready.
	// However, as an optimization, the application may call Advance while it is applying the
	// commands. For example. when the last Ready contains a snapshot, the application might take
	// a long time to apply the snapshot data. To continue receiving Ready without blocking raft
	// progress, it can call Advance before finishing applying the last ready.
	// ApplyConfChange applies a config change (previously passed to
	// ProposeConfChange) to the node. This must be called whenever a config
	// change is observed in Ready.CommittedEntries.
	// Returns an opaque non-nil ConfState protobuf which must be recorded in
	// snapshots.
	ApplyConfChange(cc pb.ConfChangeI) *pb.ConfState

	// TransferLeadership attempts to transfer leadership to the given transferee.
	TransferLeadership(ctx context.Context, lead, transferee uint64)

	// ReadIndex request a read state. The read state will be set in the ready.
	// Read state has a read index. Once the application advances further than the read
	// index, any linearizable read requests issued before the read request can be
	// processed safely. The read state will have the same rctx attached.
	ReadIndex(ctx context.Context, rctx []byte) error

	// Status returns the current status of the raft state machine.
	Status() Status
	// ReportUnreachable reports the given node is not reachable for the last send.
	ReportUnreachable(id uint64)
	// ReportSnapshot reports the status of the sent snapshot. The id is the raft ID of the follower
	// who is meant to receive the snapshot, and the status is SnapshotFinish or SnapshotFailure.
	// Calling ReportSnapshot with SnapshotFinish is a no-op. But, any failure in applying a
	// snapshot (for e.g., while streaming it from leader to follower), should be reported to the
	// leader with SnapshotFailure. When leader sends a snapshot to a follower, it pauses any raft
	// log probes until the follower can apply the snapshot and advance its state. If the follower
	// can't do that, for e.g., due to a crash, it could end up in a limbo, never getting any
	// updates from the leader. Therefore, it is crucial that the application ensures that any
	// failure in snapshot sending is caught and reported back to the leader; so it can resume raft
	// log probing in the follower.
	ReportSnapshot(id uint64, status SnapshotStatus)
	// Stop performs any necessary termination of the Node.

Node represents a node in a raft cluster.

func RestartNode

func RestartNode(c *Config) Node

RestartNode is similar to StartNode but does not take a list of peers. The current membership of the cluster will be restored from the Storage. If the caller has an existing state machine, pass in the last log index that has been applied to it; otherwise use zero.

func StartNode

func StartNode(c *Config, peers []Peer) Node

StartNode returns a new Node given configuration and a list of raft peers. It appends a ConfChangeAddNode entry for each given peer to the initial log.

Peers must not be zero length; call RestartNode in that case.

type Peer

type Peer struct {
	ID      uint64
	Context []byte

type ProgressType

type ProgressType byte

ProgressType indicates the type of replica a Progress corresponds to.

const (
	// ProgressTypePeer accompanies a Progress for a regular peer replica.
	ProgressTypePeer ProgressType = iota
	// ProgressTypeLearner accompanies a Progress for a learner replica.

type RawNode

type RawNode struct {
	// contains filtered or unexported fields

RawNode is a thread-unsafe Node. The methods of this struct correspond to the methods of Node and are described more fully there.

func NewRawNode

func NewRawNode(config *Config) (*RawNode, error)

NewRawNode instantiates a RawNode from the given configuration.

See Bootstrap() for bootstrapping an initial state; this replaces the former 'peers' argument to this method (with identical behavior). However, It is recommended that instead of calling Bootstrap, applications bootstrap their state manually by setting up a Storage that has a first index > 1 and which stores the desired ConfState as its InitialState.

func (*RawNode) Advance

func (rn *RawNode) Advance(rd Ready)

Advance notifies the RawNode that the application has applied and saved progress in the last Ready results.

func (*RawNode) ApplyConfChange

func (rn *RawNode) ApplyConfChange(cc pb.ConfChangeI) *pb.ConfState

ApplyConfChange applies a config change to the local node.

func (*RawNode) BasicStatus

func (rn *RawNode) BasicStatus() BasicStatus

BasicStatus returns a BasicStatus. Notably this does not contain the Progress map; see WithProgress for an allocation-free way to inspect it.

func (*RawNode) Bootstrap

func (rn *RawNode) Bootstrap(peers []Peer) error

Bootstrap initializes the RawNode for first use by appending configuration changes for the supplied peers. This method returns an error if the Storage is nonempty.

It is recommended that instead of calling this method, applications bootstrap their state manually by setting up a Storage that has a first index > 1 and which stores the desired ConfState as its InitialState.

func (*RawNode) Campaign

func (rn *RawNode) Campaign() error

Campaign causes this RawNode to transition to candidate state.

func (*RawNode) HasReady

func (rn *RawNode) HasReady() bool

HasReady called when RawNode user need to check if any Ready pending. Checking logic in this method should be consistent with Ready.containsUpdates().

func (*RawNode) Propose

func (rn *RawNode) Propose(data []byte) error

Propose proposes data be appended to the raft log.

func (*RawNode) ProposeConfChange

func (rn *RawNode) ProposeConfChange(cc pb.ConfChangeI) error

ProposeConfChange proposes a config change. See (Node).ProposeConfChange for details.

func (*RawNode) ReadIndex

func (rn *RawNode) ReadIndex(rctx []byte)

ReadIndex requests a read state. The read state will be set in ready. Read State has a read index. Once the application advances further than the read index, any linearizable read requests issued before the read request can be processed safely. The read state will have the same rctx attached.

func (*RawNode) Ready

func (rn *RawNode) Ready() Ready

Ready returns the outstanding work that the application needs to handle. This includes appending and applying entries or a snapshot, updating the HardState, and sending messages. The returned Ready() *must* be handled and subsequently passed back via Advance().

func (*RawNode) ReportSnapshot

func (rn *RawNode) ReportSnapshot(id uint64, status SnapshotStatus)

ReportSnapshot reports the status of the sent snapshot.

func (*RawNode) ReportUnreachable

func (rn *RawNode) ReportUnreachable(id uint64)

ReportUnreachable reports the given node is not reachable for the last send.

func (*RawNode) Status

func (rn *RawNode) Status() Status

Status returns the current status of the given group. This allocates, see BasicStatus and WithProgress for allocation-friendlier choices.

func (*RawNode) Step

func (rn *RawNode) Step(m pb.Message) error

Step advances the state machine using the given message.

func (*RawNode) Tick

func (rn *RawNode) Tick()

Tick advances the internal logical clock by a single tick.

func (*RawNode) TickQuiesced

func (rn *RawNode) TickQuiesced()

TickQuiesced advances the internal logical clock by a single tick without performing any other state machine processing. It allows the caller to avoid periodic heartbeats and elections when all of the peers in a Raft group are known to be at the same state. Expected usage is to periodically invoke Tick or TickQuiesced depending on whether the group is "active" or "quiesced".

WARNING: Be very careful about using this method as it subverts the Raft state machine. You should probably be using Tick instead.

func (*RawNode) TransferLeader

func (rn *RawNode) TransferLeader(transferee uint64)

TransferLeader tries to transfer leadership to the given transferee.

func (*RawNode) WithProgress

func (rn *RawNode) WithProgress(visitor func(id uint64, typ ProgressType, pr tracker.Progress))

WithProgress is a helper to introspect the Progress for this node and its peers.

type ReadOnlyOption

type ReadOnlyOption int
const (
	// ReadOnlySafe guarantees the linearizability of the read only request by
	// communicating with the quorum. It is the default and suggested option.
	ReadOnlySafe ReadOnlyOption = iota
	// ReadOnlyLeaseBased ensures linearizability of the read only request by
	// relying on the leader lease. It can be affected by clock drift.
	// If the clock drift is unbounded, leader might keep the lease longer than it
	// should (clock can move backward/pause without any bound). ReadIndex is not safe
	// in that case.

type ReadState

type ReadState struct {
	Index      uint64
	RequestCtx []byte

ReadState provides state for read only query. It's caller's responsibility to call ReadIndex first before getting this state from ready, it's also caller's duty to differentiate if this state is what it requests through RequestCtx, eg. given a unique id as RequestCtx

type Ready

type Ready struct {
	// The current volatile state of a Node.
	// SoftState will be nil if there is no update.
	// It is not required to consume or store SoftState.

	// The current state of a Node to be saved to stable storage BEFORE
	// Messages are sent.
	// HardState will be equal to empty state if there is no update.

	// ReadStates can be used for node to serve linearizable read requests locally
	// when its applied index is greater than the index in ReadState.
	// Note that the readState will be returned when raft receives msgReadIndex.
	// The returned is only valid for the request that requested to read.
	ReadStates []ReadState

	// Entries specifies entries to be saved to stable storage BEFORE
	// Messages are sent.
	Entries []pb.Entry

	// Snapshot specifies the snapshot to be saved to stable storage.
	Snapshot pb.Snapshot

	// CommittedEntries specifies entries to be committed to a
	// store/state-machine. These have previously been committed to stable
	// store.
	CommittedEntries []pb.Entry

	// Messages specifies outbound messages to be sent AFTER Entries are
	// committed to stable storage.
	// If it contains a MsgSnap message, the application MUST report back to raft
	// when the snapshot has been received or has failed by calling ReportSnapshot.
	Messages []pb.Message

	// MustSync indicates whether the HardState and Entries must be synchronously
	// written to disk or if an asynchronous write is permissible.
	MustSync bool

Ready encapsulates the entries and messages that are ready to read, be saved to stable storage, committed or sent to other peers. All fields in Ready are read-only.

type SnapshotStatus

type SnapshotStatus int
const (
	SnapshotFinish  SnapshotStatus = 1
	SnapshotFailure SnapshotStatus = 2

type SoftState

type SoftState struct {
	Lead      uint64 // must use atomic operations to access; keep 64-bit aligned.
	RaftState StateType

SoftState provides state that is useful for logging and debugging. The state is volatile and does not need to be persisted to the WAL.

type StateType

type StateType uint64

StateType represents the role of a node in a cluster.

const (
	StateFollower StateType = iota

Possible values for StateType.

func (StateType) MarshalJSON

func (st StateType) MarshalJSON() ([]byte, error)

func (StateType) String

func (st StateType) String() string

type Status

type Status struct {
	Config   tracker.Config
	Progress map[uint64]tracker.Progress

Status contains information about this Raft peer and its view of the system. The Progress is only populated on the leader.

func (Status) MarshalJSON

func (s Status) MarshalJSON() ([]byte, error)

MarshalJSON translates the raft status into JSON. TODO: try to simplify this by introducing ID type into raft

func (Status) String

func (s Status) String() string

type Storage

type Storage interface {

	// InitialState returns the saved HardState and ConfState information.
	InitialState() (pb.HardState, pb.ConfState, error)
	// Entries returns a slice of log entries in the range [lo,hi).
	// MaxSize limits the total size of the log entries returned, but
	// Entries returns at least one entry if any.
	Entries(lo, hi, maxSize uint64) ([]pb.Entry, error)
	// Term returns the term of entry i, which must be in the range
	// [FirstIndex()-1, LastIndex()]. The term of the entry before
	// FirstIndex is retained for matching purposes even though the
	// rest of that entry may not be available.
	Term(i uint64) (uint64, error)
	// LastIndex returns the index of the last entry in the log.
	LastIndex() (uint64, error)
	// FirstIndex returns the index of the first log entry that is
	// possibly available via Entries (older entries have been incorporated
	// into the latest Snapshot; if storage only contains the dummy entry the
	// first log entry is not available).
	FirstIndex() (uint64, error)
	// Snapshot returns the most recent snapshot.
	// If snapshot is temporarily unavailable, it should return ErrSnapshotTemporarilyUnavailable,
	// so raft state machine could know that Storage needs some time to prepare
	// snapshot and call Snapshot later.
	Snapshot() (pb.Snapshot, error)

Storage is an interface that may be implemented by the application to retrieve log entries from storage.

If any Storage method returns an error, the raft instance will become inoperable and refuse to participate in elections; the application is responsible for cleanup and recovery in this case.


Path Synopsis
Package raftpb is a generated protocol buffer package.
Package raftpb is a generated protocol buffer package.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL