Unwanted Pod Scale-down During Quiet Periods

Autoscaler scaled down too aggressively during periods of low traffic, leading to resource shortages during traffic bursts.

Find this helpful?

What Happened

Autoscaler reduced pod count during a quiet period, but didn’t scale back up quickly enough when traffic surged.

Diagnosis Steps

1Investigated autoscaler settings and found low scaleDown stabilization thresholds.
2Observed that scaling adjustments were made too aggressively.

Root Cause

Too-sensitive scale-down triggers and lack of delay in scale-down events.

Fix/Workaround

• Increased scaleDown stabilization settings to prevent rapid pod removal.
• Adjusted thresholds to delay scale-down actions.

Lessons Learned

Autoscaler should be tuned for traffic fluctuations.

How to Avoid