Back to all scenarios
Scenario #439
Scaling & Load
Kubernetes v1.23, AWS EKS

Autoscaler Skipped Scale-up Due to Incorrect Metric

Autoscaler skipped scale-up because it was using the wrong metric for scaling.

Find this helpful?
What Happened

HPA was using memory usage as the metric, but CPU usage was the actual bottleneck.

Diagnosis Steps
  • 1HPA logs showed autoscaler ignored CPU metrics in favor of memory.
  • 2Metrics confirmed high CPU usage and low memory.
Root Cause

HPA was configured to scale based on memory instead of CPU usage.

Fix/Workaround
• Reconfigured HPA to scale based on CPU metrics.
Lessons Learned

Choose the correct scaling metric for the workload.

How to Avoid
  • 1Periodically review scaling metric configurations.
  • 2Test scaling behaviors using multiple types of metrics.
Previous ScenarioNext Scenario