By manufacturer
Dell PowerEdge HPE ProLiant Lenovo ThinkSystem Fujitsu Primergy Supermicro IBM System x / Power Acer AltosHealth check: read the server before it tells you about a problem.
Systematic readout of BMC sensors, the System Event Log, disk SMART status, RAID controller logs and firmware levels. Most hardware failures are preceded by signals in the logs for days or weeks: the health check is how you intercept them before they turn into downtime.
Seven domains, one systematic checklist.
- BMC sensors: CPU, DIMM, VRM and ambient inlet temperatures; voltages; currents; fan RPM.
- System Event Log: event patterns — MCE, ECC above threshold, thermal events, PSU/fan faults.
- Disk SMART status: critical attributes, pending sectors, reallocated sectors, predictive failure, power-on hours.
- RAID controller logs: array status, BBU/Flash status, write-back enabled, historical drive drops.
- Firmware and microcode: current level vs. vendor-recommended level, open security advisories.
- OS layer status (where accessible): dmesg, Event Viewer, hardware-related errors on the operating system side.
- Security posture: TPM, secure boot, strong BMC passwords, valid certificates.
A report with prioritised actions.
The output of a health check is not "the system is OK / it is not OK": it is a report with classified action priorities:
- Urgent: conditions that require intervention within a few days (e.g. a disk in predictive failure, an exhausted BBU).
- Important: conditions to schedule over the next 1-3 months (e.g. firmware with an open security advisory, a fan drifting out of spec).
- Plannable: opportunities at 6-12 months (e.g. preventive thermal paste refresh, CMOS battery replacement).
- Informational: baseline data for future comparison.
The questions we get asked most often.
How often is a health check needed?
General rule: once a year on standard enterprise servers, twice a year on critical or heavily loaded systems. On consolidated server fleets, cyclical sampling is a workable approach.
What is the difference between a health check and continuous monitoring?
Monitoring (Zabbix, Prometheus, vendor tools such as OpenManage / SIM / XClarity Administrator) observes metrics in real time and raises alerts. A periodic health check analyses the historical series with an expert eye: slow degradation patterns that monitoring alone does not catch, correlations across different domains, proximity to critical thresholds. The two are complementary.
Does the health check require system downtime?
No, it is carried out entirely with the system powered on. BMC, SEL and SMART readout happens out-of-band or via lightweight agents. Any invasive tests (stress tests, extended memtest) are run only if the diagnosis points to a specific suspicion — and within an agreed window.