Story of How I Sold My SSD's Future

Sometimes you don't learn a lesson from a textbook or an article (you will learn something from here tho), sometimes you learn it from an accident, or rather, with a disappearing operating system.

To set the scene, I have two laptops. One is my daily driver, a machine that will run any game, any model (<20b) locally and generally refuses to complain no matter what I throw at it. The other is an old Lenovo V110, a machine so tired, but one with history who carried my spirit towards this field, but it struggles to keep VS Code open without freezing. It has one quality that mattered to me; 8GB of quality RAM and a little storage which sounded a great deal better than paying AWS every month for a tiny t3.micro instance with 1 GB of RAM and 2 vCPUs.

So I did what engineers do when they look at a pile of old hardware. I thought, "I can definitely repurpose this." The laptop was running Linux Mint, which was perfectly fine in itself, but Cinnamon alone was eating up around 2 GB of its 8 GB of RAM, and for a machine whose only job would be running my random ideas, python scripts to remind me of my laziness, and the occasional Docker container (whose purpose I'm not aware of), that felt like an unnecessary luxury. Ubuntu Server was the obvious choice. No Fancy GUI, no unnecessary services, just a terminal and whatever I chose to install on top of it. Perfectly paired with Tailscale.

As always, I flashed Ubuntu Server onto a USB drive, and here is where the story really should have ended. I had a perfectly functional second laptop sitting right there. Instead, I decided to "test" the installer on my primary laptop first, because every desktop Linux distribution I had ever used began with something like "Try or Install", where you boot into a live environment, poke around to make sure everything works, and only do the installing it for real. Mind you, a sane person would ideally do this on a VM but I rather decided to test my luck.

Ubuntu Server, as it turns out, had other ideas. Instead of a friendly desktop, it dropped me straight into the installer itself. Language, and I pressed Enter. Keyboard, and I pressed Enter. Network, and I pressed Enter. Storage, and I pressed Enter again, because everything up to that point had felt so routine that my thumb was practically on autopilot. Then it asked me to create a username for the machine, and that caught me off guard in a way none of the earlier screens had. "Isn't this usually after the disk gets formatted?" I remember thinking, pausing for maybe two whole seconds before shrugging the thought off and pressing Enter again anyway.

A few screens later, the installer froze. The interface stopped responding, and random text began overlapping itself across the screen until the whole thing dissolved into an unreadable mess, one of those glitches that makes you instinctively reach for the power button and reset the process. So I held it down, and the laptop shut off. I didn't know it at the time, but that single restart was about to turn a harmless little experiment into several days of torture for me.

The machine booted back up with only the flash drive as a boot option. Just nothing, staring back at me from an otherwise perfectly healthy-looking screen. At first I wasn't particularly worried, because bootloaders break and EFI entries disappear all the time, and I had fixed both before without much drama before. This, I told myself, was probably just another half-hour repair job.

Except it wasn't. Every command I ran pointed me towards pure self-reproach;a void that was filled with massive panic and guilt for autopiloting and contradicting everything I had ever learnt, and the worst part was I could have avoided all of this by not using the main system insert face_palm emoji x100

The moment fdisk stopped being funny

I booted into a live USB, the same one I had used for Mint a hundred times before without incident, and ran the command every Linux user reaches for the moment something feels wrong.

sudo fdisk -l

I was bracing myself for a slightly confused partition table, perhaps a missing EFI System Partition, or maybe GRUB had simply lost its marbles and gone looking for something else. Nothing a simple restart wouldn't solve but what I got instead was a single line that made my mind drop somewhere near my shoes.

Disk /dev/nvme0n1: 476.94 GiB, 512110190592 bytes, 1000215216 sectors

There were no partitions listed at all. Not "corrupted." or "unknown partition table." Just a sad little disk, completely blank and unpartitioned, a 983 GB slab of nothing, as though I had pulled a brand new SSD straight out of its retail packaging (bad joke in this economy) and plugged it in for the very first time.

I tried parted, and got the same story back. I tried testdisk next, half expecting it to sniff out the ghost of my old Cachy OS and its beautiful wallpaper (you can get it too click here↗), but all it did was scan, scan and scan, and a little more scan, and found nothing worth recovering, because there was nothing left pointing to anything. No backup GPT, no MBR signature, no leftover NTFS boot sector anywhere. The disk wasn't corrupted in the usual sense, because corruption at least leaves fingerprints behind. This was something closer to a clean wipe.

That was the moment the Ubuntu Server installer's earlier questions started replaying in my head in a far less amusing order. Language. Keyboard. Network. Storage.

Storage.

I had pressed Enter on a storage configuration screen without actually reading it, the way you click "Next" through a software installer you have seen a thousand times before and no longer bother looking at closely. Except this wasn't a "Try or Install" live environment at all. This was the real installer, and somewhere between that Enter key and the freeze, it had almost certainly already begun laying down a fresh, empty partition table on /dev/nvme0n1, my primary laptop's only drive, before the interface glitched out and I panicked and cut the power halfway through the write.

Cutting the power mid-write to a partition table is about the worst possible way to interrupt a disk operation that exists. GRUB, the Windows Boot Manager, the recovery partition, my files, none of it was technically "damaged" in the way a scratched hard drive is damaged. It simply no longer had an index telling anyone where to look. The data might, in some strict physical sense, still have been sitting somewhere, unlinked and directionless, but as far as any tool I owned was concerned, that SSD's future had already been sold off to a different operating system, and the receipt had been shredded along with the transaction. (Get the reference?)

Trying to buy back what I had sold

For the next day, I did exactly what anyone in my position would have done, which was to refuse to accept it. I pulled the HDD out of the Lenovo, the same one that had been waiting to become my server, and used it as a USB adapter to connect my dead SSD to another machine, so that I could throw heavier recovery tools at it without risking any accidental write to the drive itself.

I ran photorec across the entire drive, letting it do anything blindly in search of any signs of (life) file signatures instead of relying on a partition table that no longer existed. Twelve hours later, it had recovered exactly what you would expect from a drive that had been freshly and cleanly initialised, which is to say nothing meaningful at all. A handful of orphaned Windows processes from some Windows cache (yeah, who cares MICROSOFT?), a few stray fragments of some more windows files, and nothing at all resembling my project folders, my SSH keys, or the three half-finished side projects that I really, truly should have pushed to GitHub long before any of this happened.

I checked the SMART data next, half hoping to stumble on some heroic explanation, a failing controller perhaps, or a firmware bug, anything that would mean the data was still physically sitting there and simply unreachable. The drive reported perfect health (I guess I was simply expecting a miracle to occur atp),but it was simply doing exactly what any partitioning tool tells it to do the moment you press Enter through a screen you never actually read.

There is a very specific kind of grief in that realisation. A dying drive feels like bad luck, something that happened to you. A cleanly wiped drive that you wiped yourself, by not reading a single prompt, feels instead like a personal failing you get to relive every time you open a terminal for the rest of the week.

What actually survived

Once I stopped hoping and started counting, the honest inventory looked something like this. Anything backed up to GitHub had survived just fine, and this turned out to be most of my actual code, which remains the one genuine mercy in this entire story. Anything sitting in cloud storage or a synced folder had also survived without so much as a scratch. Local-only configuration, and one Docker Compose setup I was rather proud of and had never committed anywhere, all of that was simply gone. And the original Windows installation, along with its recovery partition and every last bit of bloatware I had spent years quietly uninstalling, was gone too, which is about the closest thing to a silver lining this whole story has to offer.

I reinstalled Windows (tiny11) fresh on the daily driver, restored what I could from GitHub and cloud sync, and rebuilt the local environment from scratch over the following weekend. It was, in the most literal sense possible, a clean start, though not one I would have chosen for myself given any say in the matter.

The Lenovo, for what it's worth, is now running Ubuntu Server exactly as originally planned, will be hosting my backend projects around the clock without a single complaint soon, having learned absolutely nothing from any of this, because it was never the machine at risk in the first place.

The actual lesson

It would be easy to end this by saying "read the installer prompts carefully," and while that is certainly true, it isn't really the lesson that matters. The real lesson has more to do with where you choose to run an experiment in the first place.

I had a spare laptop sitting six feet away, bought specifically so that I could break things on it without any consequence. I had a live USB already in my hand, built for exactly this kind of test. And yet, when the moment came to actually try something destructive, I reached for the machine with everything on it instead, simply because it happened to already be open, already logged in, already convenient.

"Try before you install" is a convention of user interface design, not a law of physics. Different installers make very different assumptions about what "try" is supposed to mean, and the only way to know for certain which kind you are dealing with is to check before you start pressing Enter on autopilot, or better still, to never let the question arise at all by testing on hardware you can genuinely afford to lose.

Will also be purchasing a Hard drive to do backups now and maybe even a little something with AI to not lose data ever again!

The SSD didn't fail me. I sold its future myself, one confident keystroke at a time, and Ubuntu Server just happened to be the one holding the pen.

also checkout this