While reconfiguring networking on that same TrueNAS storage rebuild — setting up a dedicated high-speed link so it could talk to our other servers directly, instead of sharing the regular office network — every single change we made undid itself. Not eventually. About a minute after each one.
At first that looked like a bug. It wasn’t. It was the system deliberately protecting us from ourselves.
The scariest way to lose access to a server
Anyone who’s spent real time reconfiguring network settings on a server has a version of this story: you’re connected to a machine remotely, you change an IP address or a routing setting, you hit apply — and the connection you were using to make that change is the same connection that just broke. Now you can’t reach the server to fix what you just did, because the only way in was the thing you just changed. On a physical server sitting in a rack in another room, or another building, that’s not an inconvenient afternoon. That’s a drive over, or a phone call to whoever’s on-site asking them to plug in a monitor and keyboard.
It’s such a common and avoidable way to shoot yourself in the foot that TrueNAS SCALE builds a safeguard for it directly into the operating system.
What it actually does
When you change network interface settings on TrueNAS SCALE, it applies the change immediately — but treats it as provisional. A countdown starts, and unless you explicitly confirm, from within the system, that the change worked and you still have access, it automatically reverts every interface back to the last configuration that was known to be good. No confirmation, no manual intervention needed to trigger the rollback — silence is treated as failure, and failure gets undone on its own.
That’s exactly backwards from how most software behaves, and on purpose. Normally, a change you don’t actively undo just stays. Here, a change you don’t actively confirm gets thrown away. If your new setting was fine, you click confirm and it sticks. If your new setting broke your connection — the interface came up on the wrong subnet, a VLAN tag was wrong, whatever it was — you don’t have to do anything at all. You just wait about a minute, and the system quietly puts itself back the way it was before you touched it.
We hit this because we were making real changes — standing up a dedicated 10-gigabit link on its own network segment, separate from general office traffic — and got the confirmation timing wrong more than once while getting the configuration right. Every time, instead of losing access to the box, we just watched it roll back and tried again. What could have been a frustrating, possibly-in-person recovery was instead a non-event.
This isn’t just a storage-server feature — it’s a question worth asking about anything
Not every piece of network equipment has this. A lot of routers, switches, and firewalls will happily accept a change that locks you out immediately and permanently, with no safety net at all. Whether that’s the case for a given device usually isn’t something a business owner thinks to ask about until after it’s gone wrong — and by then the answer doesn’t help.
The value of this kind of safeguard isn’t really about convenience for the person making the change. It’s about what happens on the other end of a mistake. Without it, a bad network change on equipment that isn’t easily reachable physically means a truck roll, a delayed fix, and downtime that lasts as long as it takes someone to get hands-on with the device. With it, the same mistake is a one-minute non-event that nobody outside the person making the change ever even notices.
That gap matters most for equipment that’s hard to get physical hands on quickly — a router at a remote office, a switch in a locked closet, anything managed after hours when the building’s empty. That’s exactly when a lockout is most expensive to fix and least convenient to discover.
Before your next network change, ask this
- Does your network equipment have any kind of “confirm this change or it auto-reverts” mechanism, or does a bad setting just stick until someone manually fixes it?
- If a network change locked out remote access to a piece of equipment right now, how would you actually get back in — and how long would that take?
- For equipment where hands-on access is hard (remote sites, locked closets, after-hours changes), has anyone confirmed there’s a safety net before a big change is made, not after?
The best safety features are the ones you only notice because they quietly kept a mistake from becoming a problem. This one did its job well enough that the whole incident amounted to nothing more than us trying the same setting twice.
It’s exactly the kind of hidden risk we check for on every business network we manage — not just the equipment that’s easy to reach.
Not sure if your own equipment has a safety net like this? Get in touch and we’ll help you find out before you need it.
