KubeHpaReplicasMismatch¶
Meaning¶
Horizontal Pod Autoscaler has not matched the desired number of replicas for longer than 15 minutes.
Impact¶
HPA was unable to schedule desired number of pods.
Diagnosis¶
Check why HPA was unable to scale:
- not enough nodes in the cluster
- hitting resource quotas in the cluster
- pods evicted due to pod priority
Mitigation¶
In case of cluster-autoscaler you may need to set up preemtive pod pools to ensure nodes are created on time.