# proxmox-zfs-snapshot-management-script

An interactive, **menuconfig-style** (whiptail TUI) script to manage ZFS
snapshots and related storage tasks on a **Proxmox VE** host — snapshots,
scheduled auto-snapshots, datasets, LXC bind-mountpoints, dataset replication,
and pool scrubs — all from one keyboard-driven menu.

## Features

- **Snapshots** — list, **create**, **delete**, and **roll back** ZFS snapshots
  (rollback uses `zfs rollback -r`, with a clear destructive-action warning).
- **Schedules** — cron-driven **auto-snapshots** per dataset with auto-pruning
  (keep last *N*). Includes a *"snapshot now"* action that reuses the same
  helper. Frequency presets (hourly … weekly) plus a custom cron expression.
- **Datasets** — view all ZFS datasets with `used` / `available` / mountpoint.
- **LXC Mounts** — view, **add**, and **remove** bind-mountpoints on LXC
  containers via `pct` (auto-picks the next free `mpN` slot).
- **Replication** — `zfs send | recv` a dataset to another dataset, choosing
  **full** vs **incremental** automatically based on the latest common snapshot.
- **Scrub** — start / stop a scrub and **schedule** periodic scrubs per pool.

All cron entries this tool manages live in a **single owner file**
(`/etc/cron.d/pve-zfs-tui`), rewritten in full on every change — your other
cron jobs are never touched. Two small helper scripts are installed on demand:

| Path | Purpose |
|------|---------|
| `/usr/local/bin/pve-zfs-snapshot.sh`  | snapshot + prune (run by snapshot cron) |
| `/usr/local/bin/pve-zfs-replicate.sh` | full/incremental replication (manual or cron) |
| `/etc/pve-zfs-tui/replication.tsv`    | configured replication source→target pairs |

## Requirements

- **Proxmox VE** host (Debian-based) with ZFS — provides `zfs`, `zpool`, and
  (for the LXC menu) `pct`. The non-Proxmox menus work on any ZFS-on-Linux box;
  the **LXC Mounts** menu is hidden when `pct` is absent.
- `root` — the script re-execs itself with `sudo` automatically.
- `whiptail` (installed automatically if missing).

## Usage

```bash
chmod +x zfs-snapshot.sh
./zfs-snapshot.sh
```

The main menu lets you drill into each area; every destructive action
(delete, rollback, schedule removal, mount removal) asks for confirmation,
and every command shows its output and exit code in a scrollable box.

### Scheduling an auto-snapshot

1. **Schedules → Add / update a schedule**.
2. Pick the dataset, choose a frequency (or enter a custom cron expression),
   and set **keep last N**.
3. The schedule is written to `/etc/cron.d/pve-zfs-tui`, e.g.:

   ```cron
   # Snapshot schedules (auto-prune keeps last N)
   0 3 * * * root /usr/local/bin/pve-zfs-snapshot.sh RAID1_1TB/data 7
   ```

   On each run it creates `DATASET@auto-YYYYMMDD-HHMMSS` and destroys the
   oldest `auto-*` snapshots beyond the keep limit. Manual snapshots (any name
   not starting with `auto-`) are **never** auto-pruned.

## Safety notes

- **Rollback** destroys all newer snapshots and changes since the chosen
  snapshot. The TUI warns before confirming.
- **LXC mount changes** may require the container to be stopped (Proxmox warns).
- Removing a **replication target** only forgets the pair — it does *not*
  destroy the target dataset.
- The cron file is owned entirely by this tool; edit schedules via the menus.

## License

MIT