Alerting: Scheduler to drop ticks if a rule's evaluation is too slow (#48885)

* drop ticks if evaluation of a rule is too slow.
* add metric schedule_rule_evaluations_missed_total
This commit is contained in:
Yuriy Tseretyan
2022-06-08 12:50:44 -04:00
committed by GitHub
parent ffb8ae4900
commit a89d4a5be7
7 changed files with 101 additions and 15 deletions

View File

@@ -56,6 +56,7 @@ type Scheduler struct {
SchedulableAlertRulesHash prometheus.Gauge
UpdateSchedulableAlertRulesDuration prometheus.Histogram
Ticker *legacyMetrics.Ticker
EvaluationMissed *prometheus.CounterVec
}
type MultiOrgAlertmanager struct {
@@ -199,6 +200,15 @@ func newSchedulerMetrics(r prometheus.Registerer) *Scheduler {
},
),
Ticker: legacyMetrics.NewTickerMetrics(r),
EvaluationMissed: promauto.With(r).NewCounterVec(
prometheus.CounterOpts{
Namespace: Namespace,
Subsystem: Subsystem,
Name: "schedule_rule_evaluations_missed_total",
Help: "The total number of rule evaluations missed due to a slow rule evaluation.",
},
[]string{"org", "name"},
),
}
}