Most failures on SME servers don’t arrive all at once. A data disk that is filling up, a RAID array that has been reporting “degraded” for weeks, a certificate about to expire or a backup job that quietly fails all build up until someone notices. By the time they become visible, work has usually stopped.
The purpose of regular maintenance is to catch small signs before they turn into outages. The checklist below was written for Windows Server-based SME infrastructure as daily, weekly, monthly and periodic steps; the commands run directly in PowerShell and the command line.
Why a written checklist?
Maintenance slips when it depends on one person. A check skipped in a busy week doesn’t carry over to the next; after a few months it stops being done at all. The common picture is this: everyone assumes backups are being taken, and nobody knows the date of the last successful job.
A written list gives you three things: every check is done the same way, you know who checked what and when, and the process continues when the person responsible changes.
It doesn’t need to be long. It just needs to cover the business’s truly critical systems and be applied regularly. A short list gets used. A long list stays on the shelf.
Daily checks
The daily check should take a few minutes. The goal is not deep analysis but spotting anything unusual on the same day.
- The result of overnight backup jobs: successful, warning or failed. Treat a “Warning” like a failure; a skipped VM or a locked file usually shows up there.
- Free space on system and data disks.
- Critical services running: file sharing, database and application services.
- Alerts from hardware management interfaces: disks, RAID, power supply, fans and temperature.
Disk space is the most often overlooked yet easiest indicator to monitor. The following command quickly shows how full each volume is:
Get-Volume | Where-Object DriveLetter |
Select-Object DriveLetter, FileSystemLabel,
@{Name='SizeGB'; Expression={[math]::Round($_.Size / 1GB, 1)}},
@{Name='FreeGB'; Expression={[math]::Round($_.SizeRemaining / 1GB, 1)}},
@{Name='FreePct';Expression={[math]::Round($_.SizeRemaining / $_.Size * 100, 0)}} |
Format-Table -AutoSize
The biggest consumers of system disk space are well known: WinSxS and the update cache, IIS and application logs, temporary files piling up in user profiles, and SQL Server transaction log files with no growth limit. When the C: drive fills up, services stop unexpectedly and updates fail halfway.
Weekly checks
A weekly review reveals recurring problems that don’t show up in the daily check.
Event logs
Error entries in Windows event logs are often the first sign of a failure. Grouping the last seven days of system errors by source brings the recurring ones to the top:
Get-WinEvent -FilterHashtable @{LogName='System'; Level=2; StartTime=(Get-Date).AddDays(-7)} |
Group-Object ProviderName | Sort-Object Count -Descending |
Select-Object Count, Name -First 15
Sources to look at first in the console: disk and storage drivers (e.g. disk, Ntfs, the RAID controller driver), EventLog and Kernel-Power for unexpected shutdowns, and Service Control Manager for service crashes. If the number of errors from the same source grows every week, the problem is growing.
Security and capacity
- Endpoint protection (antivirus or EDR) agents up to date and active.
- Unusual increases in failed sign-in attempts.
- Weekly trends in disk, memory and CPU usage.
- How full the backup repository is.
Monthly checks
Monthly maintenance is where planned changes and broader tests happen.
Applying updates on a schedule
Microsoft releases security updates on the second Tuesday of every month. Updates should be applied in a predefined maintenance window, not at random times.
If possible, try them on a non-critical server or in a test environment first, then roll them out to production. On ERP, accounting and industry application servers, check the vendor’s compatibility notes. A common mistake: the update is installed but the server isn’t restarted for weeks, and the “pending reboot” state causes unexpected problems at the next maintenance.
Active Directory health check
If domain controllers have problems, everything from user sign-in to file sharing is affected. Run a basic health and replication check every month:
dcdiag /q
repadmin /replsummary
dcdiag /q lists only errors; empty output is a good sign. In environments with more than one domain controller, also check replication errors with repadmin /replsummary. A DC whose replication has been failing for a long time cannot reconnect cleanly once it exceeds the tombstone lifetime, so don’t leave the error for weeks.
Other monthly steps
- A restore test for at least one critical system.
- Checking SSL/TLS certificate expiry dates.
- Reviewing unused user and computer accounts.
- Checking the list of accounts with administrator rights.
The restore test is the most often skipped item on the list. A backup isn’t a backup until it has been tested.
Quarterly and yearly checks
These checks aren’t done often, but skipping them has a big impact.
- Server firmware, BIOS and hardware management controller (iDRAC, iLO and similar) updates.
- Uninterruptible power supply (UPS) testing and battery condition.
- Tracking hardware warranty and vendor support periods.
- End-of-support dates for operating systems and software.
- Capacity planning: disk, memory and license needs for the next 12 months.
- Updating the disaster recovery plan and contact list.
A common gap on the UPS side: the UPS works, but the software or network card that sends a controlled shutdown signal to the server isn’t configured. In a long power outage the battery runs out and the server shuts down hard. In the test, try not only the battery but the shutdown scenario too.
Maintenance calendar at a glance
| Frequency | Check | Why it matters |
|---|---|---|
| Daily | Backup results, disk space, critical services | Noticing problems the same day |
| Weekly | Event logs, security agents, capacity trend | Seeing recurring errors |
| Monthly | Updates, AD health, restore test, certificates | Planned change and verification |
| Quarterly | Firmware, UPS, account and permission review | Reducing hardware and access risks |
| Yearly | Warranty, support periods, capacity and DR plan | Budget and refresh planning |
Keeping a maintenance record
The result of every check should be written down, even briefly: what was checked, what was found, what action was taken and who did it?
This record makes it easier to look back when a problem occurs and shows how the infrastructure changes over time. “The disk was 40 percent free last month, 22 percent this month” is the shortest way to explain a capacity request to management. Even a simple shared spreadsheet is far more valuable than no record at all.
When keeping the record, don’t skip two fields: the rollback step for the change made and the date of the next check. The first speeds up recovery from a bad update; the second keeps the list from being forgotten.
When to get outside support?
Maintenance doesn’t get the attention big projects do, yet it is mostly this quiet routine that keeps the business running every day. A few minutes of daily checks, planned monthly updates and periodic reviews prevent a large share of unexpected outages.
If the in-house team can’t run this calendar regularly, the list has no owner or nobody can interpret hardware alerts, it makes sense to hand the routine over to an outside team. The Eretek Bilişim Hizmetleri team runs scheduled maintenance, monitoring and update processes for SME server infrastructure in a planned and documented way.