metrics

Documentation

Index

• type Metrics
  • func New(errorRegexes []string, trackedResourceNames []v1.ResourceName, ...) (*Metrics, error)
  • func (m *Metrics) Collect(ch chan<- prometheus.Metric)
  • func (m *Metrics) Describe(ch chan<- *prometheus.Desc)
  • func (m *Metrics) DisableLeaderMetrics()
  • func (m *Metrics) EnableLeaderMetrics()
  • func (m *Metrics) LeaderMetricsEnabled() bool
  • func (m Metrics) ReportJobLeased(job *jobdb.Job)
  • func (m Metrics) ReportJobPreempted(job *jobdb.Job)
  • func (m Metrics) ReportReconcileCycleTime(cycleTime time.Duration)
  • func (m Metrics) ReportScheduleCycleTime(cycleTime time.Duration)
  • func (m Metrics) ReportSchedulerResult(result schedulerresult.SchedulerResult)
  • func (m Metrics) ReportStateTransitions(jsts []jobdb.JobStateTransitions, ...)

Types

type Metrics

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

Metrics is the top level scheduler metrics.

func New

func New(errorRegexes []string, trackedResourceNames []v1.ResourceName, jobStateMetricsResetInterval time.Duration) (*Metrics, error)

func (*Metrics) Collect

func (m *Metrics) Collect(ch chan<- prometheus.Metric)

Collect is necessary to implement the prometheus.Collector interface

func (*Metrics) Describe

func (m *Metrics) Describe(ch chan<- *prometheus.Desc)

Describe is necessary to implement the prometheus.Collector interface

func (*Metrics) DisableLeaderMetrics

func (m *Metrics) DisableLeaderMetrics()

DisableLeaderMetrics stops leader metrics from being produced. This is necessary because we only produce some metrics when we are leader in order to avoid double counting

func (*Metrics) EnableLeaderMetrics

func (m *Metrics) EnableLeaderMetrics()

EnableLeaderMetrics starts the job state and cycle metrics produced when scheduler is the leader

func (*Metrics) LeaderMetricsEnabled

func (m *Metrics) LeaderMetricsEnabled() bool

LeaderMetricsEnabled returns true if leader metrics are enabled

func (Metrics) ReportJobLeased

func (m Metrics) ReportJobLeased(job *jobdb.Job)

ReportJobLeased reports the job as being leased. This has to be reported separately because the state transition logic does work for job leased!

func (Metrics) ReportJobPreempted

func (m Metrics) ReportJobPreempted(job *jobdb.Job)

ReportJobPreempted reports the job as being preempted. This has to be reported separately because the state transition logic does work for job preempted!

func (Metrics) ReportReconcileCycleTime

func (m Metrics) ReportReconcileCycleTime(cycleTime time.Duration)

func (Metrics) ReportScheduleCycleTime

func (m Metrics) ReportScheduleCycleTime(cycleTime time.Duration)

func (Metrics) ReportSchedulerResult

func (m Metrics) ReportSchedulerResult(result schedulerresult.SchedulerResult)

func (Metrics) ReportStateTransitions

func (m Metrics) ReportStateTransitions(
	jsts []jobdb.JobStateTransitions,
	jobRunErrorsByRunId map[string]*armadaevents.Error,
)