Configuring a Storage Policy with QoS (Quality of Service) settings, including a dedicated minimum IOPS limit and a ‘High’ priority, is the most direct and effective way to guarantee performance isolation. This ensures the database VMs receive their reserved resources regardless of activity from other VMs in the same container or storage pool, directly addressing the ‘noisy neighbor’ concern.

High ‘Controller Memory’ usage specifically indicates pressure on the Nutanix Controller VM (CVM), not the AHV hypervisor or guest VM. The CVM manages all storage I/O. The first troubleshooting step should be to log into the affected CVM and check for runaway processes or unnecessary services consuming memory, as this is the most direct cause of the symptom described.

Active Directory integration configured in Prism Central does not automatically propagate to individual Prism Element instances. Each cluster’s Prism Element must be configured separately for AD authentication. The user can access Prism Central with AD credentials, but until the same AD source is configured directly on the target Prism Element, they cannot log in there.

A ‘Degraded’ Data Resiliency status for an RF2 container means one copy of the data is unavailable or stale, but the cluster has not lost data (which would be ‘Unavailable’). With all nodes up, the focus should be on why replication isn’t completing—such as a slow or interrupted rebuild process, or a persistent network partition between two nodes that is preventing synchronization, leaving the data under-replicated.

The correct and automated best practice is to use the built-in ‘Host Maintenance Mode’ feature in Prism Element. This action safely live-migrates (via vMotion-like technology) all running VMs to other hosts in the cluster, ensures the CVM is properly handled, and prepares the host for safe maintenance. It is safer and more efficient than manual migration or forced shutdowns.

Nutanix Calm is the application orchestration and automation platform designed for this exact use case. You can create a blueprint that defines the VM (software stack, network, etc.) and include application profile actions, such as a timer-set delete action. This provides a fully automated, self-service catalog item that manages the entire lifecycle, meeting the requirement with minimal admin intervention after initial setup.

A three-node cluster with RF2 (two copies of data) and ‘Node Awareness’ enabled is the standard, supported configuration for fault tolerance. Node Awareness ensures the two data replicas are always placed on different physical nodes. Therefore, if one node fails, a full copy of all data remains available on the other two nodes, allowing VMs to continue running without data loss. RF3 requires a minimum of five nodes for a failure domain of one.

AHV updates can sometimes raise the default baseline CPU model for the cluster to a newer feature set. Older VMs configured to use the cluster’s default may become incompatible if their virtual CPU is mapped to features not present on the original host CPU. The solution is to edit the VM’s settings and explicitly set its ‘CPU Model’ compatibility to the original, older model (e.g., ‘Haswell-noTSX’) to allow it to power on.

For a suspected storage performance issue, you must look at the VM’s storage I/O metrics. Consistently high IOPS and/or throughput, when combined with high latency (response time), is the direct signature of a storage bottleneck. The other metrics point to potential compute, network, or memory constraints, which are different categories of problems.

Configuring a Storage Policy with QoS (Quality of Service) settings, including a dedicated minimum IOPS limit and a ‘High’ priority, is the most direct and effective way to guarantee performance isolation. This ensures the database VMs receive their reserved resources regardless of activity from other VMs in the same container or storage pool, directly addressing the ‘noisy neighbor’ concern.0