Deploy and Configure Nutanix Clusters and Prism Central
Deployment questions test the order of operations: image the nodes, create the cluster, deploy Prism Central, register clusters, then configure categories, projects and features centrally.
Cluster bring-up
- Foundation images nodes out-of-band using IPMI and MAC addresses; plan each node's IPMI, hypervisor and CVM address before you start.
- Genesis runs on every CVM and forms the cluster; the first node initiates it and a Prism Element instance is created.
- Run NCC and fix failures before moving on, and configure DNS, NTP and the cluster VIP/VIPs.
Prism Central
- Deployed as a VM (one, or three for scale-out) with its own addresses and a VIP, then each Prism Element is registered. A PE can be registered to only one Prism Central at a time.
- Registration needs network reachability to port 9440 and valid credentials. After registration, categories, RBAC, protection policies and Flow are managed centrally.
- LCM inventory runs from Prism Central and is the supported path for AOS, AHV, PC, NCC and firmware updates.
Worked example
A new 4-node block must join an existing Prism Central managing three clusters. Order: Foundation imaging with a planned IP table → cluster creation in Prism Element → NCC run and remediation → deploy PC (large size, given the existing VM count) with three IPs plus VIP → register the new cluster → create categories such as Environment and AppType → move day-two operations to PC. Reasoning: registering before NCC passes imports a broken cluster, and creating categories before registration would leave them fragmented per cluster.
Traps
- Deploying Prism Central on the cluster it manages without reserving CPU and memory.
- Missing DNS or NTP for PC, which breaks registration and certificate validation.
- Running LCM with unresolved NCC failures or with nodes on an unsupported compatibility path.