See https://gitlab.tails.boum.org/tails/tails/-/issues/11628#note_153906. This can happen simply by applying enough automatic upgrades in a row. Doing a manual upgrade is not sufficient to fix the problem. # Symptoms There is a discrepancy between: - The disk space used on the system partition (as seen by `df`) grows after every automatic upgrade more than it should. It reaches 3GB or more. - The total size of the files stored on that partition (as seen by `du`) remains as small as expected, i.e. below 2GB. # Why this happens During shutdown, after applying an automatic upgrade, we don't cleanly unmount the system partition. So the filesystem is left dirty, with unused clusters still occupying space. # Proposed solution - To fix the root cause: try to cleanly unmount the system partition on shutdown - To recover filesystems that are already affected by the problem: on boot, repair the filesystem on the system partition and delete fsck recovery files. And then: - Users who still have enough usable free space to automatically upgrade: apply the automatic upgrade. - Users who are already unable to automatically upgrade: manually upgrade. I.e. in both cases: do what the Upgrader says. And then, the filesystem will be repaired on next boot. So no need for extra documentation or manual steps. # How users can recover in the meantime ## Fresh re-install To workaround this issue, users can: 1. Download the latest USB image 2. Install a fresh USB stick 3. Back up the old USB stick on the new USB stick: https://tails.boum.org/doc/first_steps/persistence/backup/ ## Repair the filesystem From another Linux system (which could be another Tails): 1. Repair the filesystem - Either on the command-line: run `fsck.vfat -a -w -V -v` on the affected system partition. - Or via GNOME Disks: _Repair filesystem_ on the affected system partition. 1. Mount the system partition read-write 1. Delete any `FSCK*.REC` file at the root of the system partition