Manage and Monitor a Nutanix Cluster
Monitoring questions test your ability to read Prism correctly: where an alert comes from, what is a health check versus an alert, and which tool answers which question.
Two consoles, one job
- Prism Element manages a single cluster: alerts, tasks, health, hardware, VMs, storage.
- Prism Central is the multi-cluster pane: categories, policies, projects, Calm, Flow, and cross-cluster alerts. A cluster must be registered to appear there.
Signals you must distinguish
- Alerts: something needs attention (disk failure, CVM down, capacity threshold). Acknowledging does not fix the cause.
- Events: the audit trail of what changed and who changed it.
- Tasks: long-running operations and their progress.
- NCC (Nutanix Cluster Check) runs a large set of health checks returning PASS/WARN/FAIL; it diagnoses, it does not repair.
- LCM inventories firmware/software and performs upgrades from a downloaded baseline.
Worked example
A node reports degraded. You open Prism Element, see an alert for a failed disk, confirm the disk state under hardware, check that the cluster is rebuilding rather than blocked, then run NCC to confirm no dependent check is failing. Only then do you schedule replacement. Restarting the CVM first would have destroyed the evidence and possibly the rebuild.
Traps
- Treating a green Prism dashboard as proof of health - NCC and LCM catch what dashboards miss.
- Assuming Prism Central shows a cluster that was never registered, or runs an AOS version unsupported by that PC.
- Forgetting alert routing: SMTP, SNMP traps and syslog must be configured per cluster (and per PC) to reach your monitoring stack.