OpenStack

Nova Hypervisor Resource Exhaustion Detection

critical
Resource ContentionUpdated Nov 5, 2025

Detect compute node resource exhaustion before instance provisioning failures occur by monitoring hypervisor resource metrics alongside scheduler errors.

How to detect:

Monitor the gap between total and used resources (RAM, vCPUs, disk) on hypervisors. Alert when available resources drop below thresholds or when 'No valid host found' errors appear in nova-scheduler logs alongside high resource utilization.

Recommended action:

Scale out compute capacity, rebalance workloads across hypervisors, or adjust scheduler filters. Check for scheduling inefficiencies causing uneven distribution.