Back to all scenarios
Scenario #427
Scaling & Load
Kubernetes v1.26, NGINX Ingress Controller

Insufficient Scaling Under High Ingress Traffic

Pod autoscaling didn’t trigger in time to handle high ingress traffic.

Find this helpful?
What Happened

Ingress traffic surged, but HPA didn’t trigger additional pods in time.

Diagnosis Steps
  • 1Checked HPA configuration and metrics, found that HPA was based on CPU usage, not ingress traffic.
Root Cause

Autoscaling metric didn’t account for ingress load.

Fix/Workaround
• Implemented custom metrics for Ingress traffic.
• Configured HPA to scale based on traffic load.
Lessons Learned

Use the right scaling metric for your workload.

How to Avoid
  • 1Set custom metrics like ingress traffic for autoscaling.
  • 2Regularly adjust metrics as load patterns change.
Previous ScenarioNext Scenario