System

Storage and Btrfs snapshots

SkillFishOS's safety net: automatic snapshots and rollback from boot.

One of SkillFishOS’s central ideas is being able to tinker without fear. This is made possible by the Btrfs filesystem with automatic snapshots: every important change is captured, and if something breaks you go back in one click.

Separate subvolumes

The disk uses two distinct Btrfs subvolumes:

@rootfs — the operating system;
@home — the user’s data.

Keeping them separate is essential: rolling back the system does not touch personal files. You can return to a “yesterday” system while keeping today’s documents, saves and settings.

Automatic snapshots with Snapper

SkillFishOS uses Snapper with a root configuration and pre/post hooks on APT: every time you install or upgrade packages, a snapshot is created automatically before and after. So if an update causes problems, the “before” snapshot is already there.

Configuration highlights:

a cap on the number of retained snapshots so the disk doesn’t fill up;
snapshots kept at important system milestones;
management also via the Btrfs Assistant graphical tool.

How many are kept

Five, by default: three ordinary ones (the pre/post pair around each apt operation) and two “important” ones — the upgrades that touch the kernel or systemd, which are the ones you are most likely to want back. On top of those sits the “SkillFishOS - clean install” point, which never expires: the way back to the system as it came out of the box.

The hourly timeline is off. On a home console it only eats disk without anyone ever looking at those snapshots. Snapshots you create by hand are not counted among the five and are never deleted automatically: if you made one on purpose, it stays until you remove it.

Thanks to grub-btrfs, snapshots appear directly in the GRUB menu, under “SkillFishOS snapshots”. Reboot, pick the snapshot from before the trouble, and you are inside it.

Two things worth knowing before you rely on it:

What you boot is read-only. It is a rescue environment: look around, check whether the older state really was fine, copy out the files you need. A few services will report a failure at startup — they simply cannot write. That is expected, not a fault.
The boot menu is refreshed after every apt transaction, so the snapshot taken before an upgrade is in the list exactly when you need it.

Making the return permanent

Booting a snapshot does not change anything by itself, and snapper rollback does not help here: it swaps the default subvolume, while our GRUB entry pins subvol=@ and wins. The command that does the job is:

sudo skillfish-rollback --list    # which snapshots are available
sudo skillfish-rollback 12        # snapshot 12 becomes the system, from the next boot

It moves the current system aside — it is not deleted, it becomes @-rotto-<date> — and builds a new, writable @ from the snapshot you chose, carrying the whole snapshot history across with it. If the older state turns out not to be the answer either, sudo skillfish-rollback --undo puts everything back, and --clean frees the space when you are sure.

It works from the normal system and from inside a snapshot booted read-only, which is the case that matters when the machine no longer starts.

Your home directory is never touched. @home is a separate subvolume: the system travels back in time, your files stay as they are. Handy to know, and worth remembering before you count on a rollback to bring back a document you deleted — it will not.

This is the safety net that lets even the youngest users explore the system without fear of breaking it irreversibly.

Why Btrfs and not Timeshift

SkillFishOS chose Btrfs + Snapper + grub-btrfs over solutions like Timeshift because:

the APT integration is automatic (a snapshot on every package operation);
the snapshots are native to the filesystem (instant, copy-on-write, cheap);
rollback is available from boot, even if the system no longer starts normally.