What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Configure safe HPA scale-down behavior in spec.behavior.scaleDown on an autoscaling/v2 HorizontalPodAutoscaler. Use the stabilization window to smooth brief drops in demand; add a Pods or Percent policy to limit how quickly replicas are removed. If you set both policy types, choose selectPolicy: Min to apply the smaller permitted change. Kubernetes’ default downscale stabilization window is 300 seconds, but the right policy values depend on your workload and available capacity.
Choose the control that matches the risk
Scale-down has two separate controls: a stabilization window and a rate policy. The window helps prevent a temporary metric dip from triggering a premature reduction. A policy caps the number or proportion of replicas the HPA can remove over a period. Configure both if you need smoothing and a firm limit on removal speed.
As an Amazon Associate I earn from qualifying purchases.
- Use a stabilization window when demand tends to dip briefly and you want the HPA to wait for a sustained lower recommendation.
- Use a rate policy when you need to bound how many replicas can be removed within a specified period.
- Use both when a brief dip should not cause a fast reduction and you also need a maximum rate of reduction.
Kubernetes documents configurable HPA scaling behavior as stable since v1.23, and autoscaling/v2 is the stable API version. See the Kubernetes HPA task guide and HPA concepts documentation.
Set a stabilization window
The downscale stabilization window is history-based smoothing. When considering a lower replica count, the HPA looks at recommendations during the configured interval and uses the highest recommendation in that history. This helps avoid reacting to a short-lived metric drop. Kubernetes describes its purpose as restricting replica-count flapping when scaling metrics fluctuate.
#1 Best Overall
The default window is 300 seconds (five minutes). The API accepts values from 0 through 3600 seconds; setting it to zero removes this smoothing. A longer window may suit workloads with recurring short dips or slow capacity recovery. A shorter window is appropriate only if the workload can safely shed capacity quickly and the cost and latency trade-offs are acceptable. Kubernetes does not prescribe a universal workload-specific window.
Limit the scale-down rate
Rate policies constrain replica changes over a periodSeconds interval. A Pods policy sets an absolute replica-change limit; a Percent policy sets a proportional limit. Policy values must be positive, and periodSeconds must be greater than zero and no more than 1800 seconds, according to the autoscaling/v2 API reference.
Understand selectPolicy
When multiple policies are configured, the default selectPolicy is Max, which permits the larger replica change among the policies. Set Min when you want the stricter of the listed limits. For example, combining a 10-percent-per-minute policy with a five-pods-per-minute policy and selecting Min makes the HPA follow the smaller change permitted by those policies. The precise number of pods affected by a percentage policy depends on the current replica count.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteIllustrative manifest
This example combines the default-length stabilization window with a percentage and an absolute rate cap. The Kubernetes guide documents the same policy mechanisms and example limits; they are illustrative, not a universal production recommendation or a tested configuration.
Rank #3
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: example
spec:
# scaleTargetRef, minReplicas, maxReplicas, and metrics omitted
behavior:
scaleDown:
stabilizationWindowSeconds: 300
selectPolicy: Min
policies:
- type: Percent
value: 10
periodSeconds: 60
- type: Pods
value: 5
periodSeconds: 60Size the window and limits using observed traffic patterns, startup and readiness delays, and spare capacity. A policy controls velocity; minReplicas sets the lower bound. Neither setting alone guarantees service safety.
Apply and verify the change
- Confirm the HPA uses
autoscaling/v2. Inspect the live object and verifyminReplicas,maxReplicas, metrics, andbehavior.scaleDownbefore editing. - Set the stabilization window and rate policies in the HPA manifest. Keep only values your workload can support; do not copy example limits without evaluating capacity and traffic behavior.
- Apply the updated manifest using your normal cluster deployment process, then inspect the HPA conditions, events, recommendations, and actual replica count during representative load changes.
- Check metric values and metric API availability. The HPA chooses the largest desired replica count across metrics. If one metric cannot be converted to a recommendation while another suggests scaling down, Kubernetes may skip the downscale.
- Ensure Deployment or StatefulSet manifests do not reapply a fixed
spec.replicasvalue while the HPA manages replicas. Kubernetes advises omitting that field from workload manifests in this situation because applying a fixed count can cause unwanted adjustments or flapping. - Reassess the policy after changes to traffic, metrics, startup or readiness behavior, or workload capacity.
These checks follow the controller behavior described in the Kubernetes HPA concepts documentation; observed outcomes depend on the cluster and workload.
Disable downscaling only when you mean to stop it
selectPolicy: Disabled disables scaling in the configured direction. It can serve as a temporary operational control, but the HPA will not reduce capacity while it remains in effect. If the goal is slower rather than zero downscaling, a bounded policy is generally more suitable. The behavior guide documents Disabled alongside the other selection options.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Keep scale-to-zero separate from ordinary downscale tuning
Kubernetes v1.37 documents scale-to-zero as beta for suitable HPAs using object or external metrics, not CPU or memory resource metrics. The official announcement, dated September 2, 2026, says the feature gate is enabled by default; verify the exact cluster release and control-plane configuration before relying on it. See the Kubernetes v1.37 scale-to-zero announcement.
A workload manually set to zero is not equivalent to one automatically scaled to zero: Kubernetes preserves the distinction so a manual zero can pause HPA reconciliation. Do not assume a manually zeroed workload will automatically wake in the same way as one managed by scale-to-zero behavior.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
