leader

Documentation

Index

• Constants
• Variables
• func NewLeaderLock(leaderLockDocId common.DocumentID, client *elasticsearch.Client, ...) leader.Lock
• type Conflict
  • func (n Conflict) Error() string
• type EsLock
  • func (e *EsLock) IsLeader() bool
  • func (e *EsLock) SetUTCGetter(getter func() time.Time)
  • func (e *EsLock) Start()
  • func (e *EsLock) Stop()
• type NotFound
  • func (n NotFound) Error() string

Constants

const (
	CHECKER state = iota
	PRETENDER
	USURPER
	LEADER
	STOPPED
)

Variables

var (
	IndexName = common.IndexName(".tasques_leader_locks")
)

Functions

func NewLeaderLock

func NewLeaderLock(leaderLockDocId common.DocumentID, client *elasticsearch.Client, loopInterval time.Duration, leaderReportLagTolerance time.Duration, tracer tracing.Tracer) leader.Lock

NewLeaderLock returns a new leader.Lock

Generates a random process id for the returned instance.

Types

type Conflict

type Conflict struct{}

func (Conflict) Error

func (n Conflict) Error() string

type EsLock

type EsLock struct {
	// contains filtered or unexported fields
}

* EsLock implements a very basic state machine for checking if the current `EsLock` process * is the leader based on what is stored inside the document in the leader index, as identified * by ID. * * It does this by using a simple polling mechanism along with a looping FSM that makes use of * The concurrent locking mechanisms in ES to figure out it should attempt to become the leader. * * Limitations: * - Not real time (polling..) * - Depends on not having too much drift in machine clock between all servers (time diff).

func (*EsLock) IsLeader

func (e *EsLock) IsLeader() bool

func (*EsLock) SetUTCGetter

func (e *EsLock) SetUTCGetter(getter func() time.Time)

Ignore: this is for tests

func (*EsLock) Start

func (e *EsLock) Start()

func (*EsLock) Stop

func (e *EsLock) Stop()

type NotFound

type NotFound struct{}

func (NotFound) Error

func (n NotFound) Error() string