Health Monitoring, Alerting, and Routine Maintenance
Health and maintenance questions are about noticing problems before users do and performing routine tasks without breaking anything. Expect alert handling, health scoring, call-home configuration and upgrade order.
Health surfaces
- Unisphere for PowerMax shows system health, component status, alert history and event logs (symevent). Alerts carry severities from critical down to informational; a critical alert means service-affecting or imminent risk.
- CloudIQ adds a health score (0-100) driven by performance, capacity, configuration and alert categories, plus proactive recommendations. A high score does not mean zero open alerts.
- Call Home (ConnectEMC) sends alerts and logs to Dell support. If it is misconfigured, failures get discovered by users rather than by support.
Routine tasks
- Daily: review new alerts and the drivers of the health score, confirm SRDF groups are in the expected state, confirm SnapVX policies ran, and check that failed or spare drives are rebuilding normally.
- Weekly: trend SRP consumption and front-end port utilization rather than waiting for a monthly report.
- Hardware replacement: drives, PSUs and fans are hot-swappable and redundant; verify the correct part type and let the array rebuild before removing another component.
- Configuration drift: after any change window, compare masking views, SRDF groups and snapshot policies against the expected baseline.
- Upgrades: check the compatibility matrix, upgrade Unisphere and Solutions Enabler to a release that supports the target PowerMaxOS before upgrading the array, and confirm replication and snapshot states first.
Worked example
An alert reports an SRDF link down at 03:00. Triage: confirm the alert is current rather than historical, check the RA port and its SFP, check the SRDF group state with symrdf -g query, and check the fabric for a zone or ISL change. Only then decide whether to deliberately suspend the group. Do not silence the alert first; that hides both evidence and risk.
Traps
Disabling alerts instead of resolving them; treating minor as no action; upgrading the array before the management software; and assuming the health score reflects every SRDF group's state.