Scaling¶

Use this checklist when demand is rising but replicas, nodes, or scheduling capacity are not catching up fast enough.

Main Content¶

flowchart TD
    A[Scaling symptom] --> B[Check Pending pods and HPA]
    B --> C[Check cluster autoscaler decisions]
    C --> D[Check node-pool growth lag]
    D --> E[Check storage or admission blockers]

1. Confirm whether the pressure is replica-level, node-level, or both.
2. Check for unschedulable pods before changing autoscaler limits.
3. Query cluster-autoscaler logs to see whether AKS refused to grow or never saw a valid scale-up path.
4. If pods are pending because of PVCs or webhooks, treat those blockers first.

kubectl get hpa --all-namespaces
kubectl get pods --all-namespaces --field-selector=status.phase=Pending
kubectl describe hpa <hpa-name> --namespace <namespace>
kubectl get events --all-namespaces --sort-by=.lastTimestamp
az aks show --resource-group "$RG" --name "$CLUSTER_NAME" --query "agentPoolProfiles[].{name:name,count:count,min:minCount,max:maxCount}" --output table

See Also¶

• Cluster Autoscaler Decisions
• Pending Pods
• PVC Binding Latency
• Scaling Failure

Sources¶

• Monitor AKS
• AKS monitoring data reference
• Query container logs in Azure Monitor