Kubernetes HPA underutilized info
>>>
max
(
quantile_over_time
(0.5,
kube_horizontalpodautoscaler_status_desired_replicas
[1d]) ==
kube_horizontalpodautoscaler_spec_min_replicas
) by (horizontalpodautoscaler) > 3
The rule computes the 50th‑percentile (median) of kube_horizontalpodautoscaler_status_desired_replicas over the last 1 day for each HPA and checks if that median equals the HPA’s configured spec.min_replicas. If more than three HPAs have this condition (i.e., they spend at least half the day at their minimum replica count), the alert fires.
Exporters Used
Related Alerts
Kubernetes API client errors
Kubernetes API server errors
Kubernetes API server latency
Kubernetes Container oom killer
Kubernetes CronJob suspended
Kubernetes CronJob too long
Kubernetes DaemonSet misscheduled
Kubernetes DaemonSet rollout stuck
Kubernetes Deployment generation mismatch
Kubernetes Deployment replicas mismatch
Kubernetes HPA metrics unavailability
Kubernetes HPA scale inability
Kubernetes HPA scale maximum
Kubernetes Job failed
Kubernetes Job not starting
Kubernetes Job slow completion
Kubernetes Node disk pressure
Kubernetes Node memory pressure
Kubernetes Node network unavailable
Kubernetes Node not ready
Kubernetes Node out of pod capacity
Kubernetes Node scheduling disabled
Kubernetes PersistentVolume error
Kubernetes PersistentVolumeClaim pending
Kubernetes Pod not healthy
Kubernetes ReplicaSet replicas mismatch
Kubernetes StatefulSet down
Kubernetes StatefulSet generation mismatch
Kubernetes StatefulSet replicas mismatch
Kubernetes StatefulSet update not rolled out
Kubernetes Volume full in four days
Kubernetes Volume out of disk space
Kubernetes client certificate expires next week
Kubernetes client certificate expires soon
Kubernetes pod crash looping