M7 · Observability, cost and troubleshooting
Capacity failures
49 / 61

Quota and IP exhaustion look alike until you inspect the failed resource

Both failures can leave Pods Pending while the autoscaler tries to add capacity. The decisive evidence is the failed node or address allocation event.

SignalQuota or stockIP exhaustion
Provider eventResource quota exceeded or requested machine unavailableSubnet or Pod secondary range exhausted
CheckRegional CPU, accelerator and address quotas; zone capacityPrimary node range, Pod secondary ranges and max Pods per node
FixRaise quota, change fallback shape or zoneExpand supported ranges, add discontiguous Pod CIDR or redesign density
PreventionForecast peaks and preflight quotaSize from maximum nodes with growth headroom

Compute, quota and IP space are three separate capacity budgets.