MicroNAS/README.md

# MicroNAS

Lightweight web GUI to manage SMB shares, NFS exports, and users on a single home server (Proxmox LXC/VM). Runs directly on the host as a systemd service.

## Proxmox LXC setup

Use a **privileged** Debian/Ubuntu container — required for the NFS server kernel module and quotas.

On the **Proxmox host**, load the NFS kernel module before starting the container:

```bash
modprobe nfsd
echo nfsd >> /etc/modules
```

## Install

Run everything below **inside the LXC container** as root.

### 1. System dependencies

```bash
apt update
apt install -y git samba nfs-kernel-server python3 python3-venv
```

### 2. Clone the repo

```bash
git clone https://gitea.honeyy.duckdns.org/Honey/MicroNAS.git /opt/micronas
```

### 3. Python environment

```bash
python3 -m venv /opt/micronas/venv
/opt/micronas/venv/bin/pip install flask gunicorn
```

### 4. Credentials

```bash
cat > /etc/micronas.env <<EOF
MICRONAS_USER=admin
MICRONAS_PASS=changeme
EOF
chmod 600 /etc/micronas.env
```

### 5. systemd service

```bash
cp /opt/micronas/micronas.service /etc/systemd/system/
systemctl daemon-reload
systemctl enable --now micronas
```

Open `http://<lxc-ip>:5000` in your browser.

---

## Updates

```bash
cd /opt/micronas
git pull
systemctl restart micronas
```

## Network

The service binds to `0.0.0.0:5000` — accessible from any LAN device at `http://<lxc-ip>:5000`.

For HTTPS, put nginx in front as a reverse proxy.

LAN or Tailscale only — never expose to the internet.

## Dev / test

```bash
# Quick local run (no gunicorn needed, must have flask installed):
MICRONAS_USER=admin MICRONAS_PASS=secret flask --app app.py run --debug --host=0.0.0.0
```

Note: SMB/NFS/user management calls require root and the relevant daemons — run on the actual LXC for full testing.

## Auth

All routes require a login session. Go to `http://<lxc-ip>:5000` — you will be redirected to the login page automatically. Credentials come from `MICRONAS_USER` / `MICRONAS_PASS` in `/etc/micronas.env`. If `MICRONAS_PASS` is not set the app starts unprotected and logs a warning — do not do this in production.

---

## Share settings reference

### SMB share settings

**Valid Users** — who is allowed to connect to the share. Checked by Samba before the OS even looks at folder permissions. Separate multiple entries with a space or comma. Prefix with `@` for a whole group.

```
alice bob          → only alice and bob can connect
@family            → everyone in the 'family' group
alice @family      → alice plus everyone in family
```

**Folder Permissions (Owner / Group / Mode)** — the actual Linux filesystem permissions on the shared folder. Applied by `chown`/`chmod` on the server. Valid Users controls the door; folder permissions control what's inside.

| Mode | Meaning |
|------|---------|
| `770` | Owner and group have full read/write. Everyone else: nothing. |
| `750` | Owner: read/write. Group: read-only. Others: nothing. |
| `755` | Owner: read/write. Group + others: read-only. |

**Force Group** — every file written through this share gets assigned to this group, regardless of who the connecting user is. Solves the mixed-ownership problem on shared folders: Alice writes a file, it belongs to group `family`, Bob (also in `family`) can edit it.

Typical shared folder recipe:
1. Create group `family` in the Users tab, add all users to it
2. Share: Valid Users → `@family`, Force Group → `family`
3. Folder: Group → `family`, Mode → `770`

**Force User** — more extreme version: all writes appear as one specific user. Useful when a service (Plex, Jellyfin, etc.) needs to own everything in a media folder.

**Guest OK** — allows connecting without any username or password. Anyone on the network can access the share. Good for a public read-only share (combine with a `755` folder mode so guests can read but not write). For a writable guest share, set mode `777` — but only do this on a fully trusted LAN.

**Read only** — clients can read files but not write, rename, or delete anything. Overrides writable setting.

**Recycle Bin** — when enabled, files deleted through the share go to a `.recycle` folder inside the share instead of being permanently gone. Clients (Windows Explorer, Nautilus) can browse and restore from there. Files stay until you manually delete them — there is no automatic expiry. To clean files older than 30 days add a cron job:

```bash
find /srv/yourshare/.recycle -mtime +30 -delete
```

---

### NFS export settings

NFS has no username/password concept — access is controlled by IP/hostname only (the Clients field). Once connected, the client's UID/GID is passed to the server and Linux file permissions apply normally.

**Clients** — who can mount the export. Use an IP, subnet, or `*` for everyone.

```
*                   → anyone
192.168.1.0/24      → one subnet
192.168.1.10        → one specific host
192.168.1.0/24 10.0.0.0/8   → two networks (space-separated)
```

**Read only** — mount is read-only for the client, regardless of folder permissions.

**no_root_squash** — by default, NFS maps root on the client (`uid 0`) to the anonymous `nobody` user on the server (root squashing). This prevents a compromised client from owning all server files as root. `no_root_squash` disables that mapping — root on the client stays root on the server. Useful when you manage both machines and need to run `chown`/`chmod` on mounted shares.

**all_squash** — maps every client user to the anonymous `nobody` user (or a custom `anonuid`). Nobody can impersonate their UID from the client. Simple for single-user setups where you just want consistent ownership on the server side.

---

## Mounting shares on a Linux client

Replace `<nas-ip>` with your LXC's IP address throughout.

### SMB (Samba)

```bash
# Install client tools
apt install cifs-utils

# One-time mount
mount -t cifs //<nas-ip>/sharename /mnt/nas -o username=youruser,password=yourpass

# Permanent — add to /etc/fstab:
//<nas-ip>/sharename  /mnt/nas  cifs  username=youruser,password=yourpass,_netdev  0  0
```

For better security, store credentials in a file instead of putting them in fstab:

```bash
# Create /etc/nas-credentials (root-only):
cat > /etc/nas-credentials <<EOF
username=youruser
password=yourpass
EOF
chmod 600 /etc/nas-credentials

# Then in /etc/fstab:
//<nas-ip>/sharename  /mnt/nas  cifs  credentials=/etc/nas-credentials,_netdev  0  0
```

### NFS

```bash
# Install client tools
apt install nfs-common

# One-time mount
mount -t nfs <nas-ip>:/srv/data /mnt/nas

# Permanent — add to /etc/fstab:
<nas-ip>:/srv/data  /mnt/nas  nfs  defaults,_netdev  0  0
```

`_netdev` tells the system to wait for the network before mounting — important for both SMB and NFS so the mount doesn't fail on boot.

**Multiple networks (e.g. LAN + Tailscale):** enter both subnets space-separated in the Clients field when creating the export, e.g. `192.168.1.0/24 100.64.0.0/10`. Both get the same ro/rw setting.

### Useful commands

```bash
# List available NFS exports from the server
showmount -e <nas-ip>

# Unmount
umount /mnt/nas

# Mount all fstab entries (after editing fstab)
mount -a
```