Backend: - WebDAV config managed at /etc/apache2/conf-available/micronas-webdav.conf with BEGIN/END markers per share; reload apache2 on every write - htpasswd file at /etc/micronas-webdav.passwd synced automatically on create_user, change_password, delete_user - Full CRUD: POST/PATCH/DELETE /api/share/webdav[/<name>] - _validate_webdav_users(): space-separated, no @group (not supported by Apache) - Fix: _validate_users() now allows @groupname for SMB (was incorrectly rejected) - Dashboard: apache2 status + restart button; webdav shares in /api/dashboard Frontend: - WebDAV tab: create form + table with clickable access URLs - Edit modal supports 'webdav' type (name, path, users, permissions section) - badge-webdav CSS class (green tint) in app.css - dashboard.js renders badge-<type> dynamically (smb/nfs/webdav) Docs: README WebDAV mount guide (Linux davfs2, Windows, Mac), PLANNING phase 4 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
5.9 KiB
MicroNAS — Planning
A lightweight web GUI to manage file shares on a single home server — like a minimal NAS control panel for a Proxmox LXC/VM. Manages SMB shares, NFS exports, and quotas by editing the host config files and reloading the daemons.
Design principle: keep it small. Single box, home/LAN use only, no database, no auth framework, dependency-light. If a feature needs heavy machinery, it's out of scope.
Deployment decision
Run directly on the LXC/VM as a systemd service (Flask behind gunicorn).
Not Docker.
Reason: this app's whole job is to manipulate host services (smbd, nfsd,
/etc/exports, exportfs, setquota). Containerizing it means bind-mounting
host config, running privileged, and fighting the kernel NFS server — all the
downsides of running on the host with none of the isolation benefit. Docker is
kept only as an optional dev/test sandbox.
Network: LAN-only. Never expose to the internet. Reach it via local network or Tailscale. Auth is a safety net, not an internet-facing defense.
Backups: Out of scope. Handled at the Proxmox layer (Proxmox Backup Server on a second client). This app does not implement backup.
Project structure
MicroNAS/
├── app.py Flask backend (all API routes + business logic)
├── templates/
│ ├── index.html Main SPA shell (Jinja2, no inline JS/CSS)
│ └── login.html Login form
├── static/
│ ├── css/app.css Custom dark theme (GitHub-palette CSS vars)
│ └── js/
│ ├── utils.js esc(), showToast(), apiFetch()
│ ├── dashboard.js loadDashboard()
│ ├── shares.js SMB/NFS CRUD + edit/delete/toggle
│ ├── users.js User CRUD + password modal
│ └── main.js Init + auto-refresh countdown
├── micronas.service systemd unit
├── README.md
└── PLANNING.md
Roadmap
Phase 1 — Correctness & safety ✅
- Implement
parse_smb()so/api/sharesworks. - Remove all
shell=True; use argument-listsubprocesscalls. - Validate input: names
^[A-Za-z0-9_-]+$; paths must resolve under/srv. - Session-based login (custom
/loginpage, Flask sessions, secret key derived fromMICRONAS_PASS). systemdunit +gunicorn+ README install guide.
Phase 2 — Dashboard & health ✅
- Daemon status — smbd / nfsd running? (green/red dot).
- Disk usage — free / used / total for
/srvwith progress bar. - All shares in one view — SMB + NFS, paths, access, size on disk.
- Per-share data used —
du -sbper share path. - Auto-refresh every 30 s + manual refresh button.
Phase 2.5 — UX improvements ✅
- UI redesign — GitHub-dark palette, custom CSS vars, better contrast and typography.
- Toast notifications — replaced all
alert(). - Frontend split into
templates/+static/css/+static/js/. - User management tab — list users (uid ≥ 1000), create, delete, change SMB password.
- Login page — replaces browser Basic Auth dialog.
Phase 3 — Full share management ✅
- Working delete for SMB & NFS (parse config into sections, rewrite whole file — not append-only).
- Edit existing shares — name, path, valid users (SMB); clients, ro/rw (NFS). Edit modal pre-filled with current values.
- Enable/disable a share without deleting it — SMB uses
available = no; NFS comments out the export line. - Status indicator per share row (green/red dot, reduced opacity when disabled).
Phase 3.5 — Share & dashboard polish ✅
- Permission editor in edit modal — owner, group, chmod mode per share folder. Owner/Group are dropdowns from live system users/groups.
- Group management — create/delete groups, add/remove members. Users tab.
- SMB extra settings — force user, force group, guest ok, read only, recycle bin (vfs objects = recycle, files go to .recycle inside share).
- NFS extra settings — no_root_squash, all_squash toggles.
- Server IPs on dashboard — shows all non-loopback IPv4 addresses.
- Service restart buttons — restart smbd / nfsd from the dashboard.
- Active SMB connections — live table on dashboard via smbstatus.
- Delete confirmation modal — replaces browser confirm().
- Auto-refresh countdown — 30 s timer with live seconds display.
Phase 4 — WebDAV ✅
- Apache
mod_dav+mod_dav_fsmanaged by MicroNAS. Config written to/etc/apache2/conf-available/micronas-webdav.confwith BEGIN/END markers per share;a2enconf+systemctl reload apache2. - Shares served at
http://<server>/dav/<name>(Apache on port 80). - Auth via htpasswd (
/etc/micronas-webdav.passwd) — passwords synced automatically when creating/changing user passwords in the Users tab. - Full CRUD: create, edit (name/path/users/permissions), delete.
- Dashboard: apache2 status dot + restart button.
- WebDAV badge in dashboard shares overview.
- @groupname fix in SMB valid users (was incorrectly rejected by VALID_NAME).
Phase 5 — Nice-to-have
- Simple in-browser file browser for
/srv. - Auto-create share directory on add (with correct owner/permissions).
Out of scope (intentionally)
- Backups (Proxmox Backup Server handles this).
- Multi-user accounts / roles (single admin login is enough).
- Database (config files are the source of truth).
- Internet exposure / TLS termination (LAN + Tailscale only).
- Clustering / multi-host management.
- Disk quotas — removed. Linux
setquota/repquotarequire quota support on the filesystem. The target deployment uses a ZFS-backed Proxmox LXC where this is unavailable inside the container. Per-dataset limits can be set from the Proxmox host withzfs set quota=.