Back to all scenarios
Scenario #404
Scaling & Load
Kubernetes v1.25, GKE

HPA and VPA Conflicted, Causing Flapping

HPA scaled replicas based on CPU while VPA changed pod resources dynamically, creating instability.

Find this helpful?
What Happened

HPA scaled up, VPA shrank resources → load spike → HPA scaled again.

Diagnosis Steps
  • 1Logs showed frequent pod terminations and creations.
  • 2Pod count flapped repeatedly.
Root Cause

HPA and VPA were configured on the same deployment without proper coordination.

Fix/Workaround
• Disabled VPA on workloads using HPA.
Lessons Learned

HPA and VPA should be used carefully together.

How to Avoid
  • 1Use HPA for scale-out and VPA for fixed-size workloads.
  • 2Avoid combining on the same object.
Previous ScenarioNext Scenario