Plenty of small offices, chambers and consultancies rely on one shared computer in a cupboard or a rented virtual server somewhere: it holds the files, runs the mail or the case-tracking tool, and is rarely thought about. When it fails, work stops. This guide is for the person who ended up in charge of that machine without ever being trained for it. It walks through the basic care routine known as SysOps, short for systems operations, one friendly step at a time. The subject is technology alone: read it as general orientation, not as a statement of what any law or professional rule demands of your office.
Step 1: Write down what you have
Take a sheet of paper or open a shared document. List each machine, what it does, who set it up and who answers when it misbehaves. Ask whether anyone in the office still uses a login that belonged to somebody who has left. The reference known as the Ops Atlas, which you can read in its systems operations entry, begins its first-month plan with exactly this inventory, and it is a fine starting point for a two-person office too.
Step 2: Give everyone their own key
A shared password is convenient and silent about who did what. Create a separate account for each person who needs access, and give each one only the rights they need. If a hosting provider supplies a remote-session tool, check whether it creates a shared default user with full rights; the Atlas warns that some do, and recommends per-person settings instead.
Ask whoever manages the machine to test the remote-login settings before applying them and to keep a second connection open during the change. That single habit prevents the awful moment of locking everyone out of a machine that is far away.
Step 3: Make the repeating jobs visible
Servers run quiet chores overnight: tidying temporary files, rotating logs, copying data to a backup. Ask for a list of them, with the time each last succeeded. A job that has not succeeded for a month is a finding, even if nobody noticed. Where possible, the Atlas prefers modern timers to the old scheduling table, because timers can catch up if the machine was off, can spread load, and can alert someone when a job fails.
Step 4: Watch the room left on the disk
Do not wait for a “disk full” message. Ask for a simple estimate: at the current rate of growth, how many days remain? In an illustration from the Atlas, a disk appears to have ten days left yet really has about seven, since a slice of the space is held back for the machine’s own administrator. You do not need to do the arithmetic yourself, only to ask for the answer in days each month.
Step 5: Keep a backup you have actually restored
This is the step that matters most. A copy that has never been restored is a hope, and a snapshot stored on the same disk is not a backup at all. Once a quarter, ask for a sample of folders to be restored onto a different machine and opened. Note the date and how long it took. If the process has never been tried, schedule it now.
For a firm whose work depends on documents, this is as basic as keeping the original of a signed paper in a safe place. Our own notes on checking where a court matter stands, using online tools show how much work leans on records being available when needed.
Step 6: Plan updates and restarts
Updates arrive constantly, and some only take effect after a restart. Agree a monthly slot, tell staff in advance and keep a short note of which machines are waiting for a restart. Before the planned restart, ask the person responsible to check that the disk-mount configuration is valid, since an old typo can stop a machine from starting. Also ask whether your update tool can undo a bad update; some can, and others require holding back an older version.
Step 7: Leave the safeguards switched on
When a program refuses to work, it is tempting to turn off the firewall or the extra access-control layer. Please ask your technician not to do this as a shortcut. If one rule needs relaxing, relax that rule for that program, and write down why. Disabling protections as a way of getting software to run appears on the Atlas list of habits to avoid.
Step 8: Know your first reaction to trouble
If the machine slows down or a service stops, the instinct is to restart it. Resist that. A restart erases the clues that would tell you what went wrong. First note the time, what staff reported and what the machine shows, then ask the technician to look at load, memory and disk before acting. Not every large number is a problem: a high load figure can reflect waiting on a slow disk, and low free memory is normal because the system uses spare memory as a cache.
Step 9: Test getting in the back way
Ask how the machine can be reached if the network configuration breaks. Many hosting providers offer a console that works without the network. Try it once while things are calm, and check that recovery keys for any encrypted disks are kept in a separate, safe location.
What to tell your provider or technician
If someone else looks after the machine for you, these steps double as a briefing. Ask them to show you the inventory, the list of accounts, the scheduled jobs and their last results, the disk runway figure and when a restore was last tried. A competent provider will be glad to be asked, and the answers tell you quickly whether the arrangement is solid. If the replies are vague, treat that as information in itself and agree on a plan with dates before the next busy season arrives.
A simple monthly routine
- Review the account list and remove anyone who has left.
- Read the report of scheduled jobs and their last success.
- Check the days of disk room left.
- Apply updates in the agreed slot and note pending restarts.
- Once a quarter, restore a sample from backup and time it.
If you ever need to decide quickly whether a situation needs outside help, our article on understanding when to seek immediate legal guidance covers that question for legal matters; for technology, the equivalent is a named person or provider who is on call. Write their details on the inventory sheet from Step 1.
Taken in sequence, the steps need a few hours up front and roughly an hour each month afterwards. The outcome is a machine your office can forget about, which is exactly what you want from it.