Credential Archaeology: Recovering Access to Devices Nobody Has Passwords For
Somewhere on your inherited network is a switch that has done its job flawlessly for years. It forwards traffic, survives power blips, asks for nothing. You would love to check its configuration, maybe back it up, maybe just say hello. You can't — because nobody, anywhere, knows the password. The previous administrator knew it. The previous administrator is gone.
This is the inherited network's most awkward problem: gear that works perfectly and locks you out completely. It feels urgent (it isn't, yet). It feels embarrassing (it shouldn't — it is close to universal). And the internet's advice tends to jump straight to "just factory reset it." That is how a locked-out switch becomes a down switch. There is a better way, and it works like archaeology: survey the site, dig carefully, and only bring in the heavy machinery when the careful digging is exhausted.
This guide walks that sequence. Triage which locked devices actually matter. Hunt for the credentials that almost certainly still exist somewhere. Run real password-recovery procedures without turning them into self-inflicted outages. Then comes the part that makes the whole ordeal worth it. Rebuild every recovered device into a credential system that means nobody ever has to do this dig again.
In this guide
- Locked out is not (yet) an emergency
- The dig: where old credentials hide
- Trying the obvious, without making things worse
- Recovery mode: how password recovery actually works
- Planning a recovery like the maintenance it is
- The last resort: factory reset and rebuild
- Rebuilding forward: the never-again vault
- From archaeology to architecture
Locked out is not (yet) an emergency
First, breathe. A device you cannot log into needs nothing from you to keep doing its job. It has been running without administrative attention for ages, which is precisely how you got here. Locked-out-but-running is a debt, not a fire. It becomes a fire only at the moment you need to change something. You might need to add a VLAN, patch a vulnerability, replace a failing port, or alter a firewall rule. So the goal is not "recover everything this month." The goal is to recover access in the order you are likely to need it, and to do each recovery calmly, on a schedule you chose.
That means triage. Rank your locked-out inventory by likelihood-of-needing-access-soon, not by completeness:
| Device | Why you'll need access sooner rather than later | Priority |
|---|---|---|
| Firewall / edge router | Rule-change requests arrive constantly; security fixes matter most here; it is the front door | First |
| Core switch | VLAN and port changes ripple from here; everything depends on it; you need its config backed up | First |
| Wireless controller | Wi-Fi password and network changes are routine requests | Second |
| Access switches | Port changes and troubleshooting, but each one affects fewer people | Second |
| UPS management cards | Battery alerts and shutdown integration are valuable, rarely urgent | Third |
| Printers, misc. gadgets | They fear no administrator and obey no vault; recover them last, if ever | Last |
One consolation while the list is long: admin access and monitoring access are separate doors. A device whose admin password is lost may still answer read-only SNMP queries with a community string you found in old configs. That means you can watch its health, ports, and traffic long before you can log into it. Take the read access. A locked-but-monitored device is half recovered.
The dig: where old credentials hide
Before any recovery procedure — which always means a maintenance window and usually means a reboot — comes the archaeology. Credentials are rarely gone; they are written down somewhere nobody has looked yet. Every password you dig up is a downtime window you cancel, so the hunt earns its hours. Work the sites in rough order of yield:
- Password managers and their exports. Check the predecessor's workstation for an installed password manager, and ask whether the organization ever ran a shared vault that has since been forgotten. Then search the file shares for exported credential files — people export vaults to spreadsheets "temporarily" and forget them. If you find one, treat it as the hazardous material it is: use it, then lock it away and delete the loose copies.
- Documentation fragments. The wiki's oldest pages, a OneNote notebook, a file named
network.xlsx, the predecessor's home directory, and the closing notes of old tickets ("reset admin password, documented in the usual place" — find the usual place). - The old admin workstation's browsers. Web-managed devices — firewalls, access points, UPS cards — often live on in saved browser logins. One sober sentence belongs here: examine a predecessor's machine only because it is employer-owned equipment and only with your employer's explicit authorization. This is administration of company assets, not curiosity about a person.
- Configuration backups. Hunt for TFTP folders, backup tool archives, and files that look like device configs. Configs frequently contain SNMP community strings in plain text, account names, and sometimes reusable secrets. Even a credential-free config is treasure. It is the rebuild plan you will desperately want if a reset ever becomes necessary. Archive every config you find, immediately.
- Central authentication. Check whether the "locked" devices even use local passwords. Network gear can delegate logins to a central server using protocols such as RADIUS or TACACS+. Both are ways for a device to ask "hey, authentication server, is this login valid?" A login prompt that mentions a domain, or a config line naming an auth server, changes everything. The vault you need to open is that one server. One recovered account there may unlock half the building. Check this early — it is the highest-leverage find on the list.
- Sticky notes and drawer archaeology. Yes, really. The underside of keyboards, monitor bezels, the top desk drawer, inside the rack door, and taped to the device itself. It is undignified, and it works often enough that no honest guide can omit it.
- Vendors and the MSP. Whoever sold or installed the equipment may have credentials or config archives on file. An active support contract means the vendor can walk you through recovery for your exact model. Worth an email before any reboot.
- Colleagues' memories. "It was the company name and some year" is not a password, but it is a lead — which brings us to guessing etiquette.
Trying the obvious, without making things worse
Organizations have password habits, and a few informed guesses are legitimate archaeology. Try the pattern colleagues remember, or the password that opens three other devices. Try the vendor's factory default in case nobody ever changed it. But guess like a professional, because careless guessing creates brand-new problems. Some devices lock an account or a source address after a handful of failures. Some log or alert loudly on failed logins.
Before you type a single guess: learn the device's lockout behavior. Check how many failures it tolerates, whether lockout is temporary or permanent, and whether it applies to the console as well as remote logins. If you cannot find out, assume three strikes and behave accordingly. And keep a guess log: what you tried, where, and when. It stops you from burning attempts on repeats, and it documents that your access attempts were careful, authorized administration.
A few refinements: space attempts out rather than rattling them off. Prefer trying on the least critical interface available, and stop after a few misses. The dig list above almost always has an unexplored site left. Guessing is a spice, not a strategy.
Recovery mode: how password recovery actually works
When the dig comes up dry, the good news: nearly all business-grade network equipment has a documented password recovery procedure. Vendors know admins leave, so they build in a deliberate escape hatch. They deliberately design it to require physical presence, usually a console cable and a reboot, so that a remote attacker can't use it. The details differ by vendor and model, but recovery procedures come in three general shapes. Telling them apart is the single most important skill in this article:
- Console recovery, configuration preserved. You connect to the device's console (serial) port, reboot it, and interrupt the boot sequence at the right moment. Then instruct the device to start up while ignoring its stored configuration. You log in without a password, load the stored configuration back in, set a new password, and save. The config survives; only the lock changes. This is the good path, and on most serious gear it exists.
- Reset with configuration preserved. Some devices offer a button-press ritual (a short press, or holding it during a specific boot phase) that clears only the admin credentials and keeps everything else.
- Factory reset. On other devices, the very same-looking recessed button erases everything — VLANs, addresses, rules, years of accumulated decisions — and returns the box to the day it left the factory. Same hole, different universe.
The one rule of recovery: before you press, hold, or interrupt anything, know whether the procedure preserves the configuration or erases it. Confirm this for your exact model, from the vendor's own documentation. "Recovery" and "reset" look identical from the outside, and the label on the button will not tell you. If you cannot confirm the config survives, plan the whole exercise as if it were a factory reset. Never press a recessed button on faith.
The physical toolkit is humble: the right console cable for the device, a laptop, and a terminal emulator. Older gear wants a serial cable, often the vendor's own pinout. Newer gear frequently has a USB console port. The classic serial settings, unchanged since before most of us were admins:
9600 baud, 8 data bits, no parity, 1 stop bit (8N1), no flow control
Some newer devices run their consoles faster; the manual for your model says. And this is the moment to repeat the theme quietly running through this guide. Recovering access to your employer's equipment, with your employer's authorization, is ordinary administration — not intrusion. The flip side deserves respect too. A device might belong to someone else: the managed service provider's router, the ISP's handoff box, the landlord's camera system. If so, get written confirmation of ownership and permission before touching it. One email beforehand is cheap; the conversation afterward is not.
Planning a recovery like the maintenance it is
Every recovery shape involves a reboot, and a reboot of a switch or firewall is a real outage for everyone behind it. So a password recovery is not a desk-side experiment; it is a small, plannable maintenance event, and it deserves the full ritual:
- Harvest first. Before the window, capture everything the device will give you without an admin login. Collect SNMP read data, port status, MAC tables, and the web interface's read-only pages. Take photographs of the device and every cable position. If any config backup exists anywhere, stage it where you can reach it during the window.
- Schedule and announce. Pick a time when the affected users are gone, tell people anyway, and give yourself a generous window. Recovery procedures involve reading boot messages and getting one keystroke right at one moment. Rushing them is how the good path becomes the bad one.
- Think rollback before you start. Know what you will do if the device misbehaves after the reboot: which spare covers it, which cable plan restores the old state, who gets called. Writing the back-out plan is how you find the flaw in the main plan.
- One device per window. Recovery evenings are not the time for a batch job.
- Afterward, immediately: save the configuration and back it up off the device. Store the new credential properly (next section). Verify the services behind the device actually work — not just that the lights are green.
If this cadence sounds familiar, it is the same discipline as reviewing inherited configurations: read-only first, backup before touching, changes only inside a window with a way back. The guide on auditing inherited configs without breaking anything covers that craft in depth, and a recovered device should flow straight into it.
The last resort: factory reset and rebuild
Sometimes there is no preserving path. The model's only documented procedure is a full wipe, or the recovery attempt goes sideways, or the hardware's console port turns out to be dead. Then you are in factory-reset territory. The reason it sits last on every list deserves spelling out: a reset means rebuilding the device's configuration from documentation you do not have. The VLANs, the trunk ports, the addressing, the firewall rules — all of it was in the box and nowhere else. After the wipe it is nowhere at all. Do this casually and you have converted "we can't log into the switch" into "the third floor has no network and we don't know why it ever worked."
What softens a rebuild is exactly the evidence a careful takeover produces. If you have run the discovery sprint, you hold most of the lost config's behavior in your registers. The subnet register says what networks the device served. The MAC-table dumps and port map say what hung off every port. The switch skeleton says which ports were uplinks. That is usually enough to rebuild a working configuration — not identical to the lost one, but honest and documented, which the lost one was not. Rebuild from evidence, and label the result as new.
Make the last resort gentler still: schedule it with a wider window than feels necessary. Line up vendor support beforehand if a contract exists. Practice on the least important locked device first — the closet switch, not the core. There is even a quiet upside when a rebuild is done deliberately. The device comes back with current firmware, a clean config, a known password, and actual documentation. It is frequently in better shape than at any point in its prior life. The trick is that this outcome is only available to people who chose it, planned it, and fed it with evidence. As an improvisation, the same act is just an outage you performed on yourself.
Rebuilding forward: the never-again vault
Here is the rule that turns this whole painful exercise into a one-time event: every device you recover enters the new credential regime the moment you recover it. Never re-enter a freshly recovered device into the old chaos — the shared password, the spreadsheet, the memory of whoever happened to be in the room. The new regime has four parts:
- Unique credentials per device, generated rather than invented, so that one leaked password never opens the whole building again.
- A real vault: a proper password manager with access for the people who legitimately need it — and an audit trail of who looked at what.
- Break-glass access: an offline, sealed copy of the critical credentials (and the vault's own recovery key) stored somewhere physically secure that your colleagues can find in an emergency. Its job is to make sure the vault never becomes the new single point of failure. Its job is also to make sure your someday-successor never has to write a guide like this one. Be the predecessor you wish you'd had.
- Documentation in the credential ledger you started in week one — system, address, account, where the credential lives, when access was last verified (the first-week survival guide covers starting it). The ledger holds no passwords itself; it is the map that says where the vault's keys fit.
While you are inside each recovered device — legitimately, in a planned window — spend ten extra minutes on visibility. Configure a read-only SNMP community so the device can be monitored from now on. Point its logging at your central syslog collector. That makes future authentication failures something you can see rather than discover. (Check the device's clock while you are at it — logs with wrong timestamps mislead more than they help, a trap the guide on timestamp sanity dissects.) Watched logins, monitored health, vaulted credentials: that is the never-again architecture, and every recovered device should join it on day one of its second life.
From archaeology to architecture
Seen end to end, credential archaeology is not really about passwords. It is the process by which the most opaque devices on an inherited network become its best-documented ones. Each dig ends not just with access restored but with a config archived and a credential vaulted. A break-glass copy is sealed, and monitoring watches the door. The whole arc in one table:
| Phase | What you do | What you leave behind |
|---|---|---|
| Triage | Rank locked devices by likelihood of needing access; grab read-only monitoring access where it exists | An access-priority list; watched devices |
| The dig | Hunt password managers, docs, browsers, configs, auth servers, drawers, vendors, memories; guess carefully | Recovered credentials; archived configs; a guess log |
| Recovery | Confirm recovery-vs-reset for the exact model; run it as planned maintenance with a rollback path | Access restored with config preserved and backed up |
| Rebuild (last resort) | Factory reset, scheduled generously; reconstruct from discovery-sprint evidence | A clean, current, documented device |
| Rebuild forward | Unique credentials, vault, break-glass copy, ledger entry, logging and monitoring enabled | The never-again architecture |
Work the table top to bottom, one device per maintenance window, in priority order, and the locked-out list shrinks from a source of dread into a queue of routine evenings. The rest of the takeover playbook — config audits, quick wins, and the sweep for your predecessor's other surprises — lives in Taking Over an Undocumented Network.
Put this into practice with SysPeep Monitor
No monitoring tool can dig a password out of a drawer. But the moment a device is back under your control, SysPeep Monitor can make sure it never goes dark quietly again. It auto-discovers your devices. It watches health and interfaces over SNMP, WMI, and SSH with the read-only credentials from your new vault. And it alerts you when a device stops answering. The free trial installs on one Windows machine.
Download Free TrialLearn About SysPeep