Monitoring, Alerting, and Cluster Health
Where to look
Prism Element and Prism Central expose the same core surfaces: Health (cluster-wide status), Alerts (conditions needing attention), Events (informational/past activity), Tasks (in-flight operations), and per-entity performance charts in Analysis.
Operational tooling to know:
- NCC (Nutanix Cluster Check) - the health-check framework. Run it before and after upgrades and when troubleshooting; from the CLI it is typically invoked as a full sweep of health checks.
- Pulse - automated, outbound-only telemetry to Nutanix support. It does not remediate anything; it improves support cases.
- Alert policies / email (SMTP), SNMP traps, and syslog forwarding for external monitoring integration.
Traps
- Confusing alerts with events: an alert is an active condition; an event is a record of something that happened. A resolved alert moves to history but its events remain.
- Confusing NCC (diagnostics) with LCM (lifecycle/firmware updates). They appear side by side in Prism but do completely different jobs.
- Believing Pulse is bidirectional or that disabling it breaks support entitlement workflows - it is telemetry, not a management channel.
- Missing that some checks only run on demand, so a "green" dashboard does not mean every NCC check has passed.
Rule of thumb
Something is broken or performance is degraded → start with Prism alerts and Analysis charts, then run NCC to get a structured diagnosis. Something needs updating → that is LCM, not NCC.
Worked example
After a node replacement, run NCC to confirm no failed checks and watch re-protection progress in Prism before declaring the cluster healthy - the dashboard can look green while data is still unprotected.