A disk so full, it couldn't be restored

Original Article ↗ Hacker News Discussion ↗

Filling an APFS-formatted Mac disk to capacity can leave the system unable to delete files or restore from Time Machine, with even low-level tools like `rm` failing due to “no space left on device” errors. Commenters dissect how copy‑on‑write filesystems, journaling and snapshots can deadlock when metadata also needs space, contrasting APFS’s behavior with ZFS, btrfs, ext4, NTFS and older HFS+. Many see this as a serious quality and design failure from Apple, share workarounds such as truncating large files, booting from recovery or external media, and emphasize the need for better backup strategies and reserved “slop space” to avoid bricking otherwise healthy machines.

APFS / macOS behavior with full disks

  • Multiple users report APFS volumes becoming effectively bricked when near‑100% full: rm and GUI deletion fail with “No space left on device,” even from Recovery OS.
  • The likely cause discussed: copy‑on‑write and journaling need free space for metadata/log updates; if that pool is exhausted, even deletions or truncations may fail.
  • Some say APFS is more fragile in this state than HFS+ was; others note similar issues historically with ZFS and Btrfs, which now reserve “slop space” to avoid this.
  • Several note this can also hit iPhones/iOS, causing boot loops or requiring DFU restore when storage is maxed.

Workarounds and recovery techniques

  • Suggested tactics:
    • Boot into Recovery or single‑user mode, run fsck/Disk Utility repairs on the “Data” volume, then delete large files.
    • Truncate large files instead of deleting (various shell idioms: >file, truncate -s0, :>file), though some report even truncation failing on APFS.
    • Delete APFS/Time Machine local snapshots with tmutil, or APFS snapshots via Disk Utility.
    • Clone the disk to a larger one, expand, then clean up.
    • In extreme cases: reformat and restore from backup.
  • Some describe using placeholder “ballast” files to instantly free space when a disk gets dangerously full.

Time Machine reliability and alternatives

  • Opinions are sharply divided:
    • Many describe Time Machine as chronically unreliable: sparsebundles corrupting, network restores broken (especially over SMB on recent macOS), and Time Machine volumes hard to copy or migrate.
    • Others report long‑term success over many years, especially with local SSD targets.
  • Several suggest alternatives: Backblaze (with caveats about its encryption model), Arq with cloud storage (e.g., S3/Glacier), or open‑source tools (restic/borg/kopia, rsync+ssh, ZFS).

Apple ecosystem and quality concerns

  • Commenters criticize Apple for:
    • Not preventing full‑disk states from wedging the system.
    • Declining macOS quality, especially around networking (SMB), filesystem robustness, and Sonoma regressions.
    • Pushing users toward iCloud instead of investing in robust local backup/restore.
  • Others counter that macOS and Time Machine still compare favorably to Windows or remain “good enough” for many users.