Server Maintenance Checklist: Monthly, Quarterly and Annual Tasks

Before the Checklist: Set Up Alerting

A checklist is what you do on a schedule. Alerting is what catches the failure that happens between schedules. Configure this once and half the monthly list becomes "confirm nothing alerted."

• Email alerts from iDRAC or iLO for drive, PSU, fan, memory, and thermal events, sent to an address that a person reads. Test it by pulling a PSU cable and confirming the email arrives.
• Backup job notifications on failure and, importantly, on success, so a silent stop is noticed when the daily email doesn't arrive.
• UPS monitoring software reporting battery health and any transfer to battery, and configured to shut servers down cleanly.
• An inlet temperature threshold in the management controller, set a few degrees below the ceiling of the ASHRAE recommended range, so a cooling problem is flagged before a thermal shutdown.

Monthly: About an Hour

The monthly routine is a review, not a hands-on job. Everything here is done from a browser.

Quarterly: A Morning, Partly Hands-On

The quarterly routine adds the physical checks and the things that need a maintenance window.

• Firmware review. Check BIOS, iDRAC or iLO, RAID controller, NIC, and drive firmware against current releases. Apply updates that fix security issues or known faults; skip ones that don't. Management controller firmware is the one most often neglected and the one where the vulnerabilities live.
• RAID battery or capacitor state. Not just "present" but its charge capacity and age. A learn cycle that keeps failing, or a battery past three years, goes on the replacement list before it drops the controller to write-through.
• UPS self-test. Trigger the built-in battery test and confirm it passes. This confirms the battery exists and holds charge; the annual runtime test below confirms how long.
• Physical inspection. Open the rack. Check fan noise for grinding or wobble, look for dust on intakes, confirm no cables have slipped into the exhaust path, verify both PSU LEDs are green, and check that blanking panels are still in place.
• Filter and intake cleaning. Air conditioning filters, rack door filters if fitted, and server fan intakes. In most climates this is a quarterly job. In Qatar it is not; see the section below.
• Alerting test. Deliberately trigger a non-critical alert (unplug and reconnect a PSU, or use the controller's test function) and confirm the email arrives where it should. Mailboxes change, filters get added, and alerts silently stop reaching anyone.
• Spare parts audit. Confirm the cold spare drive, PSU, and any other on-shelf spares are still there, still match the hardware in service, and haven't been borrowed for something else.
• Management credentials and access review. Who has iDRAC or iLO access, whether it's still on an isolated VLAN, and whether anyone who left still has an account.

Annually: A Planned Day

The annual routine needs a scheduled maintenance window, because some of it requires taking things offline.

• Full hardware diagnostics. Boot into the Dell Lifecycle Controller or HPE Intelligent Provisioning and run the complete diagnostic suite, including the extended memory test. This is the same test recommended in our pre-purchase inspection, and it catches faults the logs haven't recorded yet.
• UPS runtime test under real load. Pull the mains and time how long the batteries actually carry the servers. Compare to the figure from last year. Batteries lose capacity gradually and the self-test won't tell you this; only a real discharge does. If runtime has dropped below what the shutdown sequence needs, the batteries are due.
• Battery replacement decisions. UPS batteries at three years in a warm room, or five in a cool one. RAID controller batteries at three to five. Replace proactively; they don't fail loudly.
• Full restore drill. Restore a complete server or VM to alternate hardware and confirm it boots and the application works. This is the test that proves the backup strategy, not the monthly file restore.
• Thorough physical clean. Servers out on rails, lids off, compressed air through heatsinks, fans, and PSUs. Rack cleaned, cable management tidied, blanking panels reinstated.
• Capacity and lifecycle review. Storage growth over the year, CPU and memory headroom, drive ages and SSD wear trends, and which hardware is approaching the end of its useful life. This is when next year's purchases get planned rather than forced.
• Documentation refresh. Network diagram, IP allocations, service accounts, licence keys and renewal dates, and the recovery procedure. The document nobody updates is the one nobody can use during an outage.

Adjusting the Schedule for Qatar

The intervals above are the industry defaults, and they assume a temperate, clean environment. Two local factors pull several items forward.

• Dust: monthly, not quarterly. Shamal conditions load air conditioning filters and server intakes far faster than manufacturer schedules assume. Move filter checks to monthly, and do a visual check of server intakes at the same time. After a significant dust event, check regardless of schedule.
• Batteries: shorter cycles. Lead-acid battery life roughly halves for every 10°C above 25°C. In any room that runs warm, plan UPS battery replacement at three years rather than five, and RAID batteries at three rather than four or five. Verify with the annual runtime test.
• Temperature trends: watch May to September. Inlet and drive temperatures that were comfortable in February will climb through the summer. The monthly temperature check matters most in the months before the heat peaks, so cooling problems are found before July rather than during it.
• Air conditioning service before summer. Have the server room cooling serviced in spring, not when it fails in August. If there's a second unit for redundancy, confirm both run and both can carry the load alone.

What to Automate and What Needs a Person

The distinction matters because automation catches events and people catch trends. A drive logging one correctable error triggers nothing. A drive logging one a week for three months is about to fail, and only someone looking at the log over time sees that.

The One-Page Version

Frequently Asked Questions

How often should I update server firmware?

Review quarterly, apply when there's a reason. Security fixes and known-fault corrections justify an update; a release with no relevant changes doesn't. Management controller firmware (iDRAC, iLO) deserves the most attention because that's where remotely exploitable vulnerabilities appear. Always update during a planned window with a verified backup.

What is the most important maintenance task?

Testing restores. Everything else on the list reduces the chance of a failure; a verified restore is what saves the business when one happens anyway. A backup that has never been restored is an assumption, not a backup. Restore a file monthly and a whole server annually.

How do I know when to replace a UPS battery?

Not from the self-test, which only confirms the battery holds some charge. Run a real runtime test annually by pulling mains under load and timing it. When runtime drops below what the shutdown sequence needs, or the batteries reach three years in a warm room or five in a cool one, replace them. In Qatar's heat, plan on the shorter figure.

Do I need to clean inside the server?

Yes, at least annually, and more often in dusty environments. Dust on heatsinks and fan blades insulates components and raises temperatures even when the room is cool. In Qatar, check intakes monthly and do a full clean with compressed air annually or after any significant dust event.

Can maintenance be fully automated?

The monitoring can. The judgement can't. Automation catches events: a drive failing, a temperature crossing a threshold. A person catches trends: a drive that's been logging errors for months, a runtime figure that's shorter than last year. Automate everything measurable and keep a human on the monthly log review and the annual tests.

Where can I buy replacement server parts in Qatar?

ServerDove stocks RAID batteries, power supplies, drives, fans, and UPS batteries for Dell PowerEdge and HPE ProLiant servers in Doha, with same-week delivery across Qatar and help confirming the correct part from your server model and log entry.

Final Thoughts

Enterprise servers are built to warn you. Every component that can wear out reports its state, and every controller logs what happened. Maintenance is mostly the discipline of reading what the hardware is already saying, on a schedule, and replacing the consumables before they turn a warning into an outage.

An hour a month, a morning a quarter, a day a year. Set the alerting, keep the spares, test the restores, and adjust the intervals for the heat. Servers maintained this way retire on schedule rather than at 2am.