allochealth

Documentation

Index

• Constants
• type Tracker
  • func NewTracker(parentCtx context.Context, logger hclog.Logger, alloc *structs.Allocation, ...) *Tracker
  • func (t *Tracker) AllocStoppedCh() <-chan struct{}
  • func (t *Tracker) HealthyCh() <-chan bool
  • func (t *Tracker) Start()
  • func (t *Tracker) TaskEvents() map[string]*structs.TaskEvent

Constants

const (
	// AllocHealthEventSource is the source used for emitting task events
	AllocHealthEventSource = "Alloc Unhealthy"
)

Types

type Tracker

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

Tracker tracks the health of an allocation and makes health events watchable via channels.

func NewTracker

func NewTracker(
	parentCtx context.Context,
	logger hclog.Logger,
	alloc *structs.Allocation,
	allocUpdates *cstructs.AllocListener,
	consulClient serviceregistration.Handler,
	checkStore checkstore.Shim,
	minHealthyTime time.Duration,
	useChecks bool,
) *Tracker

NewTracker returns a health tracker for the given allocation.

Depending on job configuration, an allocation's health takes into consideration - An alloc listener - Consul checks (via consul API) - Nomad checks (via client state)

func (*Tracker) AllocStoppedCh

func (t *Tracker) AllocStoppedCh() <-chan struct{}

AllocStoppedCh returns a channel that will be fired if the allocation is stopped. This means that health will not be set.

func (*Tracker) HealthyCh

func (t *Tracker) HealthyCh() <-chan bool

HealthyCh returns a channel that will emit a boolean indicating the health of the allocation.

func (*Tracker) Start

func (t *Tracker) Start()

Start starts the watcher.

func (*Tracker) TaskEvents

func (t *Tracker) TaskEvents() map[string]*structs.TaskEvent

TaskEvents returns a map of events by task. This should only be called after health has been determined. Only tasks that have contributed to the allocation being unhealthy will have an event.