Taking a Real Vacation as a Department of One
Ask a solo admin why they have not taken a proper vacation in two years and you will hear something about dedication, or timing, or "it's just easier not to." The honest answer underneath is usually dread. You dread the phone lighting up on a beach, what will be waiting on return, or being the reason payroll did not run. That dread is not a personality flaw. It is what happens when a system has exactly one operator and no designed failover. The fix is the same as for any other availability problem: you engineer it.
This guide treats your two weeks off exactly the way you would treat a planned outage of a critical server, because that is what it is. You will build the coverage plan (most things can wait; a short list cannot). You will build the break-glass document that makes a non-IT colleague useful in a crisis. You will make the monitoring changes that keep the noise away from your beach. And you will build the structured re-entry that replaces the traditional first-day-back panic. And if working through it reveals that the business genuinely cannot survive two weeks without you, that finding is valuable too. We will talk about what to do with that finding.
In this guide
- What actually needs covering: the honest triage
- Shrinking the risk before you leave
- Who covers, at solo scale
- The sealed envelope: your break-glass pack
- Vacation mode for your monitoring
- The re-entry protocol
- If two weeks away genuinely risks the business
- The vacation-prep checklist
- Where to go next
What actually needs covering: the honest triage
Start by writing down everything you do in a normal two-week stretch, then sort it with one brutal question: what actually breaks if this waits until I am back? The exercise is liberating, because the honest answer for most of the list is "nothing." New laptop setups wait. The projector in the small conference room waits. Software requests, report tweaks, the printer that only jams for one person — all of it waits, and the world does not end. Users are far more tolerant of "IT is on vacation, back on the 15th" than solo admins believe, provided someone told them in advance and the truly urgent stuff has a path.
That leaves the short list: things that cannot wait two weeks. For a typical small shop it is genuinely short. It includes outage response for the systems everyone depends on, plus account lockouts and password resets (small, but they stop a person's whole workday). It includes anything payroll-adjacent, because payroll happens on a schedule that does not care about your flights. For each item on the short list you have exactly three tools, applied in this order of preference. First, prevent it from coming up at all. Second, delegate it to a named person with instructions. Third, defer it with a clear promise. Here is how that sorts out in practice:
| Duty | Can it wait two weeks? | Plan |
|---|---|---|
| Outage of a critical system | No | Delegate: designated colleague or MSP + break-glass pack + monitoring routed to them |
| Password resets and lockouts | Not really | Prevent: self-service reset enabled beforehand. Delegate the leftovers to a colleague with reset rights only |
| Payroll-adjacent systems | No | Prevent: verify the full run works the week before; brief the vendor's support line; delegate the "who to call" card |
| Backups running | No — but needs a watcher, not a fixer | Delegate the daily "did the green email arrive?" glance |
| New hires / onboarding | Yes, with planning | Defer: prepare accounts before you go, or schedule starts around the trip |
| Purchases, projects, requests | Yes | Defer: announce the freeze window and the return date |
| The misc queue (printers, peripherals…) | Yes | Defer: tickets accumulate; you triage on return |
Notice how little actually requires a human with your skills during the window. The whole plan exists to protect perhaps five scenarios — which is what makes it buildable.
Shrinking the risk before you leave
The cheapest coverage is for problems that never happen. So the week before you leave is a quiet period, run with the same discipline as a change freeze before a big launch. No changes that week: no firmware updates, no new GPOs, no "quick" firewall rule cleanups, nothing whose failure mode is "weird behavior three days later". In other words, change nothing at all. Most incidents follow changes; a network nobody touched is a network that mostly keeps working. Patches deserve special mention: get them applied and settled — deployed, rebooted, observed for a few normal days — at least a week out. Do not apply them freshly the day before you fly.
Then spend the quiet week verifying rather than changing. Confirm backups are succeeding and do one real restore test. A backup you have restored from is insurance; one you have not is a rumor. Confirm the UPS batteries are healthy and disk space has months of headroom rather than weeks. Confirm certificates are nowhere near expiry and no DHCP scope is flirting with full. Your monitoring history makes this a twenty-minute review rather than a scavenger hunt. You are simply reading the trends and asking "what is most likely to go wrong in the next two weeks?" Then you defuse whatever candidates you find, while defusing is still easy.
Who covers, at solo scale
The designated colleague
You probably cannot hand the pager to another admin, because there isn't one. But a capable, calm, non-IT colleague armed with excellent instructions beats an empty chair by a wide margin. Think of the operations manager who is good with checklists, not the person who "knows computers." Their job is not to fix things. It is to execute the runbook, communicate, and escalate. They power-cycle the thing the instructions say to power-cycle, tell people what is happening, and call the right outside number when the script runs out. Choose them weeks ahead, get their (and their manager's) genuine agreement, and walk them through the break-glass pack in person for half an hour. The walkthrough matters more than the pack; it converts a scary document into "oh, this is just a checklist."
The MSP retainer
A managed service provider (MSP) is a company that does IT operations for many clients. An MSP will often take a short-term or on-call arrangement covering a vacation window. For a business of any real size, this is the professional answer for the outage scenario. Arrange a named contact and phone path, plus a briefing call the week before. The MSP should see your network map and know your critical systems by name. Agree response expectations for the window, and arrange temporary, scoped access. Create accounts for them, with the permissions the likely scenarios need, enabled for the window, disabled on your return. Never hand over the domain-admin keys casually to cover a beach week; access you can grant, you can also scope.
Tier zero: the vendor support numbers
Below both options sits the layer solo admins forget they already pay for: support contracts. Your ISP, your firewall vendor, your backup vendor, the payroll system's support desk — each will talk a calm delegate through their own product's crisis. Make a one-page list of who to call, with account numbers and contract IDs filled in. It turns "everything is down and Sam is in Portugal" into a set of phone calls a non-technical person can actually make. It is the cheapest coverage on this page.
The sealed envelope: your break-glass pack
The break-glass pack is the document that makes all of the above work: what your delegate opens when something breaks. ("Sealed envelope" is the tradition; a restricted folder they can reach plus one printed copy in a locked drawer is the practical version. It must survive the network being down. A wiki page on the dead file server is a joke the universe will eventually make at your expense.) Five sections:
- How to reach emergency credentials. One sober paragraph here, because this is where people do something regrettable: do not print passwords. A printed password list in a drawer is a breach waiting patiently. Instead, use your password vault's emergency-access feature, or a single break-glass account. That account's credential is sealed separately, every use is logged, and you rotate its password the day you are back. The pack describes the path to credentials and who may authorize opening it — not the credentials themselves.
- The five-most-likely-problems runbook. Short, numbered, decision-light instructions for your top realistic scenarios: internet down, the main server unresponsive, Wi-Fi down, the line-of-business app down, power outage and recovery. Write them for a smart non-expert: exact names, exact locations, photos of the right rack, "the switch labeled CORE, third from the top." The craft of writing procedures a stranger can follow at 2 AM is its own discipline. The runbooks guides cover it. But for the pack, aim lower and clearer: power-cycle this, wait ten minutes, then call the number in section three.
- Who to call. List the MSP if you have one, the ISP with account and circuit numbers, key vendors with contract numbers, and you. Include an honest line about when: "call me if the outage affects the whole company and the MSP has not responded within an hour."
- The do-not-touch list. Every environment has its haunted machinery: the fragile legacy box, the server that must never be power-cycled without a ritual. Name them and say plainly: if the problem seems to live here, skip the runbook and go straight to the phone numbers.
- Where to look. Include the monitoring console's address and a read-only login. The delegate can then answer "is it everything, or just one thing?" before touching anything. That is the single most useful question in any outage.
Packs rot, so tie the refresh to the trip. The week before you leave, reread every page and test the vault's emergency-access path end to end. Dial-check the two most important numbers and update whatever changed since last time. Thirty minutes, calendar-scheduled, same appointment as the delegate walkthrough.
Tip: the pack you build for vacations is also your bus-factor insurance — the same document serves if you are ever out unexpectedly. That reframing helps with motivation: you are not doing paperwork so you can go to the beach. You are closing the biggest single risk in the company's IT, and the beach is the deadline.
Vacation mode for your monitoring
If you built the three-lane routing — page-worthy, morning digest, weekly summary — vacation mode is a temporary rewiring of the lanes, not a rebuild. The changes, as a checklist:
- Route the page lane — the true-outage alerts — to the delegate and/or the MSP for the window. They receive only what the short list covers; they should not see a disk-space warning in their lives.
- Point the digest and weekly lanes at a folder, not your phone. They will be the backbone of your return sweep; nobody needs them during the window.
- Keep the heartbeat alive: the monitoring system's own "I am running" signal should go somewhere a human will notice its absence — the delegate's inbox is fine. A monitoring system that dies silently on day two turns your whole plan into decoration.
- Establish one channel for escalation to you, and give it a distinct sound. If the delegate or MSP truly needs you, they use that channel, and anything on it is real. Everything else about your phone gets to be a camera and a map for two weeks.
- Suppress what the quiet period makes irrelevant — and write down every routing change you make, because you will be undoing each one on return.
- Force a test alert through the new routing before you leave, and watch it land on the delegate's phone.
And the anti-pattern, named plainly: leaving all alerts flowing to your own phone "just to keep an eye on things." That is not a vacation; that is remote work with sand. It also quietly insults the plan you just built. If the routing, the delegate, and the pack are real, your attention is redundant. If they are not real, fix them rather than substituting yourself. The point of the machinery is that you can stop being the machinery, and two weeks of genuinely not watching is both the reward and the test.
The re-entry protocol
The traditional first day back — opening an inbox of six hundred messages and excavating in chronological order — is how a rested admin gets un-rested by 10 AM. Replace archaeology with a structured sweep. Block the first ninety minutes, decline everything, and work this sequence:
- 1. What fired? Read the alert history for the whole window, resolved included. You are reconstructing the fortnight's story: what broke, what recovered on its own, what the delegate handled.
- 2. What was suppressed or digested? Sweep the accumulated digests and weekly summaries for slow movers that crept while you were away. Look for disk trends, certificate countdowns, DHCP scopes, anything now closer to a threshold than you would like.
- 3. Did backups run the whole time? Verify every day of the window, and spot-restore one file. Two silent weeks is exactly how backup failures grow teeth.
- 4. What changed? Debrief the delegate and read the MSP's notes: what they touched, what they worked around, what users reported. Their workarounds are your repair list.
- 5. Restore normal routing. Undo every vacation change, using the list you wrote. This is the step everyone forgets, discovered three weeks later when a real alert goes to the wrong phone.
- 6. Close the loop. Thank the delegate concretely, then spend ten minutes updating the pack while it is fresh: what confused them, what was missing, what number went unanswered. Next year's prep just got shorter.
Only then open the inbox — and triage it knowing the systems' side of the story already, which defuses most of its drama. The queue of deferred requests is real work, but it is Tuesday work, not emergency work.
If two weeks away genuinely risks the business
Sometimes building this plan surfaces an uncomfortable truth: there is no plausible delegate and no budget for an MSP week. In that case, there is a genuine chance that two quiet weeks end in a crisis nobody present can handle. Hear this clearly: that is not your failing, and it is not fixed by you never resting. A business that cannot tolerate one employee's absence for two weeks has a resilience problem. You skipping vacations does not close it — it hides it. Meanwhile, the underlying single point of failure (you) accumulates fatigue and, eventually, a resignation letter or a hospital visit. Flu and jury duty do not submit vacation requests.
What the failed vacation test produces is evidence — and evidence is exactly what changes minds. Write it down in business terms: "for two weeks in July, an internet outage would have had no qualified responder; estimated impact, the whole company offline for days." Pair it with the operational record you are already keeping — the uptime evidence file of incidents and near-misses. Together, they become the opening exhibit in the conversation about coverage, retainers, or a second pair of hands. That conversation has its own guide: Knowing When You Need Help. Take the vacation anyway, with the best plan you can assemble; a stressed fortnight with an imperfect plan still beats another cancelled one.
The vacation-prep checklist
| When | Do |
|---|---|
| T-2 weeks | Confirm the delegate and their manager's blessing; book the MSP window if using one; apply patches so they settle; announce the freeze and your dates to the company; start the do-nothing-new discipline |
| T-1 week | Quiet period: no changes. Refresh the break-glass pack; test vault emergency access; restore-test a backup; review monitoring trends for brewing problems; walk the delegate through the pack in person; brief the MSP |
| Day before | Switch alert routing to vacation mode and send a test alert to the delegate; verify last night's backup; set out-of-office replies pointing to the delegate for emergencies and the return date for everything else; write down every routing change |
| While away | Nothing, unless the escalation channel sounds. That channel stays on; everything else stays off |
| Return, first 90 min | The six-step sweep: alert history, digests, backups, delegate debrief, restore routing, update the pack and say thank you |
Where to go next
The first engineered vacation is the hardest; the second one reuses the pack, the delegate, and the checklist, and mostly just needs dates. What you have really built is proof that the operation can survive without your constant presence. That same proof underpins a sustainable pace the other fifty weeks of the year. It is also the strongest card in your hand when the time comes to ask for help. The rest of The IT Department of One continues there.
Put this into practice with SysPeep Monitor
A vacation plan leans hard on monitoring you can trust while you are not looking. SysPeep Monitor's alert rules and notifications let you route the true emergencies to a delegate for two weeks while summaries wait quietly in a folder. Its dashboards and network maps give a stand-in or an MSP a fast, readable picture of a network they have never seen. Try it on your own network with the free trial.
Download Free TrialLearn About SysPeep