ASG (Aldertech Storage Governor)

"Why does df -h say I have 2.6TB free but BTRFS thinks it's 3.5TB?"

Because df is wrong — and nobody else fixes this.

ASG (Aldertech Storage Governor) is a lightweight, zero-dependency Python tool that provides accurate health monitoring, capacity planning, and intelligent scrub scheduling for BTRFS RAID1 pools — especially those with mismatched drive sizes.

If you've ever mixed a 4TB drive with two 1TB drives and wondered why your free space numbers don't add up, ASG is for you.

---

The Problem

BTRFS RAID1 with mismatched drives is powerful but poorly served by standard tools:

• df -h is mathematically wrong. It doesn't account for RAID1 pairing constraints on mismatched drives.
• btrfs scrub is fire-and-forget. The standard approach (btrfs scrub start -Bd) blocks indefinitely with zero awareness of system load — your media server stutters while the scrub runs.
• Metadata concentration is invisible. BTRFS may silently place all metadata chunks on your two smallest drives. If either fails, the entire pool is unrecoverable. No standard tool warns you about this.
• No capacity prediction. Without historical tracking, you won't know your pool is filling up until it's too late.

The Solution

ASG provides five integrated governance modules:

1. Real Capacity Engine — Calculates true RAID1 free space using the formula min(sum/2, sum - largest), tracks historical usage, and predicts days-to-full.
2. Heuristic Scrub Controller — Runs scrubs in the background, monitoring /proc/loadavg and /proc/diskstats. Automatically pauses when your system is busy and resumes when idle.
3. Integrity Monitor — Polls btrfs device stats for all drives, detects errors, and sends alerts with deduplication.
4. Metadata Concentration Watchdog — Warns if metadata/system chunks are dangerously concentrated on a subset of drives.
5. Snapshot Bridge — Creates atomic read-only snapshots before your backup software runs, with automatic cleanup.

---

Features

• Zero dependencies — Python standard library only (PyYAML optional for config)
• Accurate RAID1 maths for 2, 3, 4, or more mismatched drives
• Load-aware scrub throttling using /proc/loadavg and /proc/diskstats
• Metadata concentration alerts — catches the silent killer
• Days-to-full prediction from historical usage data
• Pre-backup atomic snapshots with retention management
• Pool UUID validation before every operation (prevents wrong-pool disasters)
• --dry-run on all filesystem-modifying commands
• Notifications — Discord, NTFY, and Gotify (all optional)
• Config-driven — single YAML file, sensible defaults

---

Requirements

Requirement	Minimum	Notes
Linux kernel	4.0+	Any kernel with BTRFS support
btrfs-progs	5.0+	sudo apt install btrfs-progs
Python	3.9+
BTRFS profile	RAID1	RAID1C3/RAID1C4 should also work
sudo	Required	BTRFS commands require root
PyYAML	Optional	For config file; falls back to defaults without it

---

Quick Start

Option 1: Install via pip

pip3 install aldertech-asg

Option 2: Install via Git Clone

# 1. Clone the repository
git clone https://github.com/jaldertech/asg.git
cd asg

# 2. Run the installer (requires root)
sudo bash setup.sh

# 3. Edit your config
sudo nano /etc/asg/config.yaml

# 4. Test it
asg status

Option 3: Run directly (no install)

git clone https://github.com/jaldertech/asg.git
cd asg
python3 -m asg status

---

Configuration

Copy config.yaml to /etc/asg/config.yaml and edit it:

pool:
  mount: "/mnt/my_pool"
  # uuid: "optional-pin-for-safety"

scrub:
  load_threshold: 3.0           # Pause scrub above this 1-min load average
  io_threshold_percent: 40.0    # Pause if any drive exceeds this I/O %

capacity:
  chunk_fullness_warn_percent: 90.0
  free_space_warn_percent: 10.0
  metadata_min_device_count: 3  # Warn if metadata on fewer devices

snapshots:
  directory: ".snapshots"       # Relative to pool mount
  retention_days: 3

# Optional — remove if you don't want notifications
notifications:
  discord:
    webhook_url: "https://discord.com/api/webhooks/..."

If PyYAML is not installed, the tool runs with built-in defaults (mount point: /mnt/media_pool).

---

Usage

# Full status overview (integrity + capacity + snapshots)
asg status

# Real capacity report with days-to-full prediction
asg capacity

# Run a load-aware scrub (pauses during high I/O)
asg scrub --dry-run     # Preview first
asg scrub               # Run for real

# Check all drives for errors
asg check

# Pre-backup snapshot (cleanup old + create new)
asg snapshot --dry-run
asg snapshot

# Clean up expired snapshots only
asg cleanup

# Use a custom config path
asg --config /path/to/config.yaml status

Example Output: Capacity Report

ASG Capacity Report — /mnt/media_pool
Timestamp: 2026-03-07 19:45:57
=================================================================

  Total Raw Capacity:   7454.1 GiB
  Largest Drive:        3727.4 GiB
  RAID1 Usable Ceiling: 3726.7 GiB
  Data Used:            100.5 GiB
  Real Free Space:      3625.8 GiB
  Utilisation:          2.7%
  Days to Full:         ~342

  Per-Device Allocation:
  Device           Data       Meta     System      Unalloc      Total
  ---------------------------------------------------------------
  /dev/sdc         103.0G     1.000G     0.031G      1761.3G    1863.7G
  /dev/sdd          30.0G     1.000G     0.008G       900.5G     931.5G
  /dev/sde          25.0G     1.000G     0.008G       905.5G     931.5G
  /dev/sdf          50.0G     1.000G     0.031G      3676.2G    3727.4G

  WARNINGS:
    ! Data chunks are 96.6% full — new chunk allocation imminent.
=================================================================

---

Cron Integration

Add to your crontab (crontab -e):

# Daily capacity report at 00:20
20 0 * * * asg capacity >> ~/asg.log 2>&1

# Weekly throttled scrub on Sunday at 02:00
0 2 * * 0 asg scrub >> ~/asg.log 2>&1

# Daily pre-backup snapshot at 02:00
0 2 * * * asg snapshot >> ~/asg.log 2>&1

---

How the RAID1 Maths Works

Standard df divides total raw space by the data ratio (2.0 for RAID1), giving sum / 2. But this ignores the pairing constraint: each RAID1 chunk must fit on two different drives.

For mismatched drives, the real limit is:

usable = min(sum_of_all_drives / 2, sum_of_all_drives - largest_drive)

Example: 4TB + 2TB + 1TB + 1TB = 8TB raw

Calculation	Result	Why
sum / 2	4 TB	Standard RAID1 formula
sum - largest	4 TB	The 4TB drive can only mirror 4TB of data across the other 3 drives
Usable ceiling	4 TB	min(4, 4) — balanced in this case

But with 8TB + 1TB + 1TB = 10TB raw:

Calculation	Result	Why
sum / 2	5 TB	Looks generous
sum - largest	2 TB	The 8TB drive can only mirror 2TB across the others
Usable ceiling	2 TB	6TB of the 8TB drive is wasted

This tool calculates the correct number automatically.

---

Verified Hardware

Hardware	OS	Pool Layout	Status
Raspberry Pi 5 (16GB)	Raspberry Pi OS Bookworm	4TB + 2TB + 1TB + 1TB RAID1	Verified

ASG uses generic Linux interfaces (/proc, btrfs-progs). It should work on any Linux system with BTRFS. If you test on other hardware, please open a PR to add it to this table.

---

Notifications

All backends are optional and independent. Configure any combination in config.yaml:

Backend	Config Key	Notes
Discord	notifications.discord.webhook_url	Standard Discord webhook
NTFY	notifications.ntfy.url	Full topic URL
Gotify	notifications.gotify.url + .token	Base URL + app token

---

Security

ASG runs btrfs commands via sudo. It requires that the executing user has passwordless sudo for btrfs (standard on Raspberry Pi OS) or is run as root.

The tool:

• Never writes to arbitrary paths (only state files in a configured directory)
• Validates the pool UUID before every operation
• Supports --dry-run for all destructive commands
• Has zero network dependencies (notifications are opt-in)

---

Related Projects

• ADRG — Aldertech Dynamic Resource Governor: Automatic CPU/RAM management for high-density Docker stacks.

Licence

MIT — see LICENCE file.

---

Built by Aldertech