Containerd Space Recovery Status | Generated by AI

Home 2026.06 English 中文

---

Containerd rsync killed, partial data cleaned up. Models move is done and working. Status:

Root disk:  697G  88G used (610G free)  — was 124G, freed 36G
Volume:     100G  37G used (58G free)

What worked:

• ✓ llama.cpp/models (37G) → volume, symlinked back

What didn’t:

• ✗ containerd (59G) — 39,000+ tiny files in overlayfs snapshots (ROCm container layers). rsync would take forever.

containerd options:

1. systemctl stop containerd && mv /var/lib/containerd /mnt/volume_.../containerd && ln -s /mnt/volume_.../containerd /var/lib/containerd && systemctl start containerd — mv is instant on same filesystem… but these are different filesystems so it’d copy too.
2. Prune unused container images instead — crictl rmi --prune could reclaim significant space without moving anything.
3. Change containerd’s root in /etc/containerd/config.toml to point at the volume (requires restart, might break running containers).

Want me to try pruning container images first? That’s the safest way to free space on root without risking the containerd setup.

---

Back Donate