Your First Week With an Inherited Network: A Survival Guide
It is day one. Your badge works, and your chair was adjusted by someone with very different ideas about lumbar support. Somewhere behind the drywall hums a network you are now responsible for. The previous administrator is gone — new job, retirement, or "no longer with the company," delivered in a tone that discourages follow-up questions. The documentation is a wiki with four pages, three of them about a phone system the company replaced.
And here is the strange part: everything works. Email flows. People log in. The printers print, mostly. That is genuinely good news — and it is also a trap. A working network with no documentation is a system whose reasons have left the building. Every cable, every config line, every mystery service is doing something, and nobody can tell you what. The moment you tidy something up "because it obviously does nothing," you find out exactly what it did.
This guide is your first week, one day at a time. By Friday you will know what you can log into, whether your backups are real, who knows what, and what the closets actually contain. Thanks to a small set of monitoring checks, you will hear about the next failure before your users do. Best of all, almost none of it requires changing a single setting.
In this guide
- The survival mindset: learn, don't improve
- The notes discipline that makes it all stick
- Day 1: Access triage and the backup panic button
- Day 2: The human map
- Day 3: The physical walk-through
- Day 4: First monitoring
- Day 5: One honest document
- What not to do in week one
- The week-one checklist
The survival mindset: learn, don't improve
Your instincts will scream at you this week. You will find passwords taped to monitors and a switch stack cabled like a plate of spaghetti. You will find a server named TEMP-DO-NOT-USE that is very clearly in production. Every professional bone in your body will want to fix all of it. Don't. Your job this week is to learn the network, not to improve it. Improvement comes soon, and it comes far faster and safer once you know what you are improving.
Here is the prime directive of an inherited network: everything currently running is load-bearing until proven otherwise. The weird service on the aging server? Load-bearing. The static route that makes no sense? Load-bearing. The unlabeled cable that disappears into the ceiling? Assume it is holding up the building. A network that has run for years without documentation survives on an accumulated pile of small, invisible decisions. On day one you cannot tell the decisions from the debris.
So start a "don't touch yet" list, and put these on it from hour one:
- Configs. Read them as much as you like; change nothing — no cleanup, no "obvious" fixes, no commenting things out to see what happens.
- Weird services. A process you don't recognize on a server you don't understand stays running. Note it, research it, leave it.
- Unlabeled cables. Photograph them, trace them with your eyes, wonder about them — do not unplug them.
- Anything with uptime measured in years. Do not reboot it. Long uptime is evidence of stability and a warning that nobody knows whether it survives a restart.
- Scheduled tasks and scripts. That 2 AM batch job is somebody's payroll feed. It runs until you know what it feeds.
None of this is forever. It is just for now. The motto of week one: look, listen, and write everything down. That brings us to the one habit that turns this week into a foundation.
The notes discipline that makes it all stick
Before the daily plan, adopt the tool that makes the plan pay off for years: an append-only daily log. One plain text file. Every entry gets a date and time. New lines go at the bottom. You never rewrite or delete old entries — that is the "append-only" part. It is not a polished document; polish comes on Friday. It is a lab notebook, and it looks like this:
=== Tuesday (day 2) ===
09:10 firewall admin login WORKS (creds from drawer note; added to ledger)
09:40 backup job NIGHTLY-FULL last succeeded 11 days ago (!!)
10:15 Maria in accounting: "the file server acts up every month-end"
11:02 core switch fan is making a noise fans should not make
-> added to the fear list
13:30 found unmanaged 5-port switch under desk in sales. photo taken.
Why so strict about appending? Three reasons. First, your memory of this week will decay in days, and detail you don't capture now is gone. Second, in three months someone will ask "when did that start?" You will be the only person in the building with an answer that has a date on it. Third, it quietly protects you. "The fan was already making that noise on my second day, and I have notes" is a sentence you may one day be very glad to say. Future-you will thank present-you, repeatedly.
Day 1: Access triage and the backup panic button
First mission: find out what you can actually log into. Sit down with whatever you were given — ideally an admin account, possibly a sticky note. Work through the list, one system at a time. Check the domain admin account, the firewall's management page, and the switches (SSH or web interface). Check the virtualization host, the backup console, the wireless controller, the UPS management card, and the ISP account portal. You are testing keys today, not redecorating: log in, confirm it works, look around gently, log out.
As you go, start the credential ledger — a simple table you will maintain for the rest of your tenure. Record system, address, account name, where the credential is stored, and date last verified. Every login you confirm goes in. Every login you lack goes in too, marked "no access." That second list matters just as much. Gear that works perfectly while locking you out completely is its own project — and, importantly, not an urgent one. The guide on credential archaeology covers recovering those devices safely when the time comes; for today, just record them.
Where do day-one credentials come from? In descending order of dignity, start with a password manager (jackpot — ask who holds the master key). Next comes an "IT passwords" spreadsheet on a share, then the predecessor's notebook. Last is the drawer of sticky notes, which is real and has saved more admins than anyone admits. Gather everything into the ledger. Do not rotate any passwords yet — that rule has its own horror story later in this guide.
Then the panic-button basics: backups. Find the backup system and answer four questions. What is being backed up? Where do the copies go? When did a job last succeed — not run, succeed? And can you actually restore? Don't take the console's word for it: restore one small file to an alternate location today and open it. A backup nobody has restored from is a hope, not a backup.
The one true emergency: if the last good backup is weeks old, or nothing restores, that outranks everything else in this article. Every other risk in your first week is survivable because restores work; a working restore is the safety net under the whole takeover. If the net is missing, drop the schedule and fix backups first.
Day 2: The human map
The documentation did not vanish entirely — it is distributed across the heads of everyone who works here. Day 2 is for interviews. Not meetings: short, friendly, fifteen-minute chats with a notebook. People love being asked what they know, and what they know is remarkable.
Who to talk to: your manager (what does the business actually expect from IT, and what burned them before?). Talk to whoever fielded complaints when the last admin was out and the longest-tenured employee you can find. Talk to one power user per department and the office manager or receptionist. Include any outside parties you can identify: the ISP contact, a managed service provider, "the consultant who did the phones." Ask each of them some version of:
- What breaks often? What do you just live with because reporting it never helped?
- What is fragile — the thing everyone knows not to touch because it once took days to revive?
- What did the previous admin spend their time on? What did they complain about?
- When something breaks, whom do you call — officially and actually?
- What runs on a schedule that must not fail — payroll, month-end, the Monday reports?
- Do you know of any binders, drawers, or folders of IT stuff anywhere?
Do not skip the front desk. The receptionist knows things the wiki never did. They know which closet the "internet guy" always asks to see and who holds the alarm code. They know that the ceiling leaked over the wiring closet two winters ago. Treat everything you hear as valuable and unverified — write it down, mark it as rumor, and confirm it later. By the end of the day you should have three artifacts: a fragile-things list, a sketch of who-calls-whom, and a set of rumors to check. All three go in the log.
Day 3: The physical walk-through
Time to meet the hardware. Get keys to every space with blinking lights: the server room, every wiring closet, the storage room that "has some IT stuff in it." Walk them all. Yesterday's interviews will have told you where the keys live and which doors exist; the office manager is usually the keeper of both.
This is the phone-camera documentation blitz, and the rule is simple: photograph everything. Rack fronts and rack backs. Every device label, model number, and serial. Patch panels and the runs into them. The UPS nameplate and, crucially, what is plugged into the UPS versus straight into the wall. The spot where the internet physically enters the building (the ISP's handoff box, called the demarcation point). Wide shots first for context, then close-ups. Don't sort or rename anything now — one folder per room, dated, dumped straight from the phone. Photos are the fastest documentation on earth: searchable later, wrong never.
While you are in each room, note in the log:
- Every device: make, model, serial, and what its indicator lights are doing
- Power: what is on battery-backed outlets, what isn't, and whether the UPS itself looks healthy
- Labels — record them all, but trust them the way you trust rumors (verify later)
- Unlabeled cables: photograph the visible run; do not unplug or "reseat" anything
- Environment: temperature, dust, water stains, and any space heater sharing a power strip with a rack (you will find one eventually)
The walk-through almost always produces at least one surprise: an unmanaged switch under a desk, a consumer-grade router doing something unexplained, a server nobody mentioned in any interview. Perfect. Into the log they go, with photos. Finding surprises in week one, on foot, is infinitely better than finding them in month six, during an outage.
Day 4: First monitoring
Monitoring is the one improvement allowed in week one, because done right it is read-only. It watches, asks devices polite questions, and changes nothing about how the network behaves. And it transforms your position. Right now, your alerting system is a user walking to your desk. Even ten basic checks make tomorrow strictly safer than today. The next failure announces itself to you first — quietly, with a timestamp — instead of arriving as a hallway ambush.
Don't build a cathedral today. Set up pings against the firewall, the core switch, and the handful of servers everyone depends on. Add disk-space checks on those servers. Add a check on backup job results, so Day 1's discovery can never quietly recur. That is a morning's work with almost any monitoring tool. The guide on the first ten checks to set up walks the full starter list in priority order, with sensible starting thresholds for each.
Day 4 also starts two clocks quietly running in your favor. The first is baseline data: from today, "normal" is being recorded — response times, traffic, disk growth. So in a month, "is this normal?" becomes a question with an answer. The second is evidence: when someone tells you "the network is down all the time," you will have graphs. Sometimes the graphs vindicate the network; sometimes they vindicate the complainer. Either way, you stop arguing from vibes.
Day 5: One honest document
Friday is for consolidation. Take the week — the log, the interviews, the photos, the credential ledger, the first alerts — and write one document with three headings. What I know: verified facts, each with its evidence ("backups restore — tested Tuesday"; "core switch is a 48-port model X — photo"). What I don't know: the honest inventory of gaps — subnets you can't explain, boxes you can't identify, devices you can't log into, whether the switch configs are even saved. What scares me: the fear list — the eleven-day backup gap you fixed, the fan making that noise, the single unlabeled cable that apparently feeds the warehouse.
The fear list is not a diary entry; it is next month's work plan wearing an honest name. Every fear converts into a task. Unknown topology becomes next week's discovery sprint. Locked-out devices become credential archaeology. Expiring certificates and end-of-life gear become a landmine sweep. Ranked fears become a plan; a plan beats a vague sense of dread every single time.
Then share a short version with your manager. Here is what I verified, here is what I don't know yet, and here is what I am doing about it, in order. It feels vulnerable to present a document that is one-third confessed ignorance. Do it anyway. Nobody expects you to know the network in a week. But everybody remembers the admin who could say precisely what they didn't know and what they'd do about it. Honest status plus a plan builds more trust than confident hand-waving ever will.
What not to do in week one
Every rule in this section was learned by some admin, somewhere, the hard way. Each one comes with the story it prevents.
No cleanups
The classic: a new admin tidies the rat's nest in the closet and unplugs a cable that "clearly went nowhere." At 8 AM they learn that it fed the barcode scanners in the warehouse — via an unmanaged switch balanced on a ceiling tile. Cosmetic cleanup has real outage potential and zero week-one payoff. The spaghetti has waited years; it can wait until you have a map.
No upgrades
Firmware updates matter and their day will come — after you have configuration backups and a maintenance window. The story here is the aging switch that took an update, rebooted, and came back blank, taking with it a configuration that existed nowhere else on earth. In week one you don't yet know whether configs are backed up (many inherited networks: no), which makes any upgrade a coin flip you don't need to take. Auditing and archiving the configs comes first; upgrades after.
No password rotations yet
It feels like security hygiene. It is a minefield, twice over. First: old service-account passwords are wired into forgotten places — scheduled tasks, the scanner that emails PDFs, an application's config file. Rotating the account breaks them all at once, mysteriously. Second: rotating device passwords before your credential ledger is complete and tested is how you lock yourself out and become your own predecessor by Thursday. Rotation absolutely happens — deliberately, later, one system at a time, with a vault behind it.
No reboots for tidiness
A reboot is a test you haven't prepared for. Elderly hardware sometimes fails to spin back up; more often, the danger is configuration that exists only in running memory.
The classic landmine: on a lot of network gear, the running configuration and the saved startup configuration can differ. A switch that has been up for three years may be running three years of changes that were never saved. Reboot it, and it wakes up in the distant past, or as a factory-fresh blank. Before any deliberate reboot of network equipment, ever, confirm the running config is saved and archived somewhere off the device.
Notice the pattern: every "don't" is really a "not yet." Cleanups, upgrades, rotations, and reboots all become safe — routine, even — once discovery, config archives, and a credential vault exist. The point of week one's restraint is to make week five's improvements boring.
The week-one checklist
| Day | Mission | You are done when... |
|---|---|---|
| Day 1 | Access triage + backup verification | Credential ledger started (including the "no access" list), and one test restore has actually succeeded |
| Day 2 | Interview the humans | Fragile-things list, who-calls-whom sketch, and rumor list exist in writing |
| Day 3 | Physical walk-through | Every closet visited and photographed; hardware notes in the log; surprises recorded |
| Day 4 | First monitoring | About ten read-only checks are running and can reach you when something breaks |
| Day 5 | Consolidate | One document — know / don't know / scares me — written and shared |
| All week | The append-only log | You wrote it down. All of it. With dates. |
Five days, zero changes, and you now hold more verified knowledge about this network than existed anywhere when you arrived. Next comes the fun part: turning "what is actually out there?" from a fear into a map. That is a structured week of its own — the discovery sprint — and it builds directly on the access, notes, and monitoring you just set up. The rest of the series lives in Taking Over an Undocumented Network.
Put this into practice with SysPeep Monitor
Day 4 is the easiest day to start early. SysPeep Monitor installs on a single Windows machine, auto-discovers the devices and interfaces you just inherited, and starts watching them over SNMP, WMI, and SSH. Its alert rules are designed to stay quiet until something is genuinely wrong. The free trial fits comfortably inside week one.
Download Free TrialLearn About SysPeep