Supportability and impact of reducing or removing resource reservations on NSX Manager
Article ID: 452983
Updated On:
Products
Issue/Introduction
This article provides the official technical guidance regarding the reduction or removal of CPU and memory reservations for VMware NSX Manager appliances. It outlines the risks to management plane stability, database integrity, and the supportability policy for environments, including VMware Cloud Foundation (VCF), where resource constraints may be encountered during ESXi host maintenance windows.
Symptoms: Resource constraints in the management cluster during host evacuation or maintenance mode may lead to requests to adjust reservations. Failure to maintain mandatory reservations can result in:
- NSX Manager UI unresponsiveness or API request timeouts.
- CorfuDB (distributed database) desynchronization or split-brain conditions.
- Control plane latency affecting security and routing policy updates.
- "InsufficientResourcesFault" when attempting to power on or migrate virtual machines.
- Appliance service crashes due to memory ballooning or hypervisor swapping.
Environment
- VMware NSX-T Data Center
- VMware NSX 4.x
- VMware vSphere ESXi 7.0, 8.0
Resolution
Reducing, or removing resource reservations on the NSX Manager virtual machine is strictly unsupported. The 100% memory reservation is a mandatory requirement designed to guarantee dedicated physical resources for latency-sensitive databases. To manage resource pressure during maintenance, implement the following supported alternatives:
- Stagger Workload Evacuation: Manually vMotion non-essential or smaller workloads individually prior to placing a host into Maintenance Mode to avoid concurrent resource spikes.
- Adjust Non-Critical VM Reservations: Temporarily lower memory reservations or power off non-critical infrastructure virtual machines (such as secondary logging, analytics, or monitoring tools) that are not required for core management plane functionality.
- Review Cluster Sizing: If utilization consistently reaches 95% during N-1 operations (where N is the number of hosts), the cluster lacks sufficient N+1 headroom. Expand the host cluster capacity to restore safe operational margins.
- Resource Pool Priority: Ensure NSX Manager appliances are placed in a high-priority resource pool to prevent contention from non-management workloads.
Additional Information
Feedback
