Skip to content

Storage Management

The #1 problem in homelabs: mixed drive sizes. You bought a 2TB drive two years ago, an 8TB last year, and a 12TB last week. Traditional RAID wastes space by sizing everything to the smallest drive. HomelabARR PE solves this with SnapRAID + MergerFS + an intelligent Cache Mover.

Architecture

Storage Tier Architecture
├── Cache Tier (NVMe/SSD) — Fast storage for active data
│   ├── /mnt/cache/appdata/      # Docker databases (PERMANENT)
│   ├── /mnt/cache/downloads/    # Active downloads (temporary)
│   ├── /mnt/cache/media/        # New media files (ages to array)
│   └── /mnt/cache/system/       # OS and binaries (PERMANENT)
│
├── Array Storage (SnapRAID + MergerFS) — Bulk media storage
│   ├── Mixed drive sizes supported (2TB + 4TB + 8TB + 12TB)
│   ├── Dual parity protection (up to 2 drive failures)
│   └── Unified namespace via MergerFS at /mnt/storage
│
└── Cache Mover — Intelligent file aging
    ├── Moves aged files from cache to array automatically
    ├── Protects Docker databases (never moved)
    └── Configurable age thresholds per directory

How Mixed Drives Work

Traditional RAID with 2TB + 4TB + 8TB + 12TB = 8TB usable (everything sized to smallest).

HomelabARR PE with the same drives:

Drive Size Role
Drive 1 2TB Data
Drive 2 4TB Data
Drive 3 8TB Data
Drive 4 12TB Data
Parity 1 12TB+ Parity (must be >= largest data drive)

Usable space: 26TB (all data drives combined). Parity protects against any single drive failure.

With dual parity, add a second parity drive for protection against two simultaneous failures.

SnapRAID Configuration

SnapRAID provides parity-based protection for your data drives. Unlike real-time RAID, SnapRAID syncs on a schedule — perfect for media libraries that change infrequently.

# Managed via the PE dashboard or CLI
storage:
  snapraid:
    parity:
      - /mnt/parity1
      - /mnt/parity2    # Optional dual parity
    data:
      d1: /mnt/disk1
      d2: /mnt/disk2
      d3: /mnt/disk3
      d4: /mnt/disk4
    content:
      - /mnt/disk1/.snapraid.content
      - /mnt/disk2/.snapraid.content
    exclude:
      - "*.unrecoverable"
      - "/tmp/"
      - "/lost+found/"
    schedule:
      sync: "0 4 * * *"     # Daily at 4 AM
      scrub: "0 5 * * 0"    # Weekly Sunday at 5 AM

Operations

Operation What It Does When
Sync Updates parity to match current data Daily (scheduled)
Scrub Verifies data integrity against parity Weekly (scheduled)
Check Reports any data corruption On demand
Fix Rebuilds corrupted files from parity On demand

MergerFS Pool

MergerFS creates a unified view of all your data drives at /mnt/storage. To your applications, it looks like one massive drive.

storage:
  mergerfs:
    mount: /mnt/storage
    drives:
      - /mnt/disk1
      - /mnt/disk2
      - /mnt/disk3
      - /mnt/disk4
    policy: epmfs    # Existing path, most free space
    options:
      - cache.files=partial
      - dropcacheonclose=true
      - category.create=epmfs

Placement Policies

Policy Behavior Best For
epmfs Existing path, most free space General use (default)
mfs Most free space Even distribution
lfs Least free space Fill drives sequentially
rand Random Testing

Cache Mover

The Cache Mover automatically ages files from fast NVMe/SSD cache to the slower array based on configurable rules.

storage:
  cache_mover:
    cache_path: /mnt/cache
    array_path: /mnt/storage
    schedule: "*/30 * * * *"    # Check every 30 minutes
    rules:
      - path: /mnt/cache/media/
        max_age: 24h             # Move media after 24 hours
      - path: /mnt/cache/downloads/
        max_age: 1h              # Move completed downloads after 1 hour
      - path: /mnt/cache/backup/
        max_age: 7d              # Move backups after 7 days
    exclude:
      - /mnt/cache/appdata/      # Docker databases stay on cache FOREVER
      - /mnt/cache/system/       # System files stay on cache
      - "*.part"                 # Incomplete downloads protected

Why Cache Matters

Operation NVMe Cache HDD Array
Database access 3.5 GB/s 150 MB/s
Active downloads 1.8 GB/s 100 MB/s
New media ingest 1.8 GB/s 100 MB/s
Aged media playback — 400 MB/s (sufficient for 4K)

Plex metadata, Sonarr/Radarr databases, and qBittorrent temp files live on NVMe permanently. Media files start on NVMe for fast processing, then age to the array where they stream just fine.

Dashboard Integration

The PE dashboard provides real-time visibility into your storage:

  • Pool overview — Total capacity, used space, drive health
  • Drive status — Individual drive temperatures, SMART data, I/O
  • Cache utilization — What's on cache, what's aging out, age thresholds
  • SnapRAID status — Last sync time, scrub results, any errors
  • MergerFS view — Unified pool with per-drive breakdown

Next Steps