HPA Scale-in Stabilization Window

HPA uses the maximum recommended replica count from the configured stabilization window. This parameter dampens short-lived load dips, eliminating needless pod churn.

For details, see the official documentation.

Value Range

Valid time format strings

Default Value

5m0s

Suggestions

Raise the window to stop rapid scale-in/scale-out caused by transient dips. This improves system stability.

Possible Risks

If the stabilization window is too short, HPA will over-react to brief load dips and repeatedly create and delete pods. If it is too long, HPA will keep unneeded pods alive after real demand has dropped, wasting resources. If it exceeds the maximum length supported (153,722,867 minutes), HPA may fail to start. When HPA restarts, the scale-in stabilization timer resets.