Hello,

I'm planning to migrate my ~100TB Windows Server 2019 (with DrivePool) to Unraid and would appreciate some feedback on my migration strategy.

I used Claude AI to create a migration plan, but I'd like to get it reviewed by the community to make sure I'm not missing anything or if there are any outdated practices.

My setup goals:

No parity drive - mostly media files that are easily replaceable

512GB M.2 SSD - cache drive for Sabnzbd extraction and Docker databases

4TB WD Purple - dedicated to Frigate (security camera footage)

2x 8TB WD Red - documents and photos with file duplication/mirroring

Remaining drives (~9 drives) - simple media storage for Plex (no protection)

Does this approach make sense, or are there better practices I should consider? Any feedback appreciated!

---

Your Drive Setup Summary

Total: ~106TB raw capacity

Unraid target configuration:

• 🟣 WD Purple (4TB) - Frigate only

• 🔴 2x WD Red (8TB) - Documents/photos with duplication

• 📦 9x Mixed drives - Media storage (no protection)

• 💾 SSD 512GB - Cache drive for databases/Docker

---

Migration Strategy: "Rolling Consolidation"

Since you have no parity, this is actually easier! We'll consolidate data onto larger drives first, then build Unraid array.

---

Phase 1: Preparation (Still in Windows)

Goal: Consolidate all data onto 8-9 drives, leaving 4-5 drives empty

Step 1: Calculate your actual data

Move all data TO these drives (keep full):

1. ✅ 12TB WD (Drive 1)

2. ✅ 10TB WD (Drive 2) 

3. ✅ 18TB WD (Drive 5)

4. ✅ 8TB SMR (Drive 3)

5. ✅ 8TB SMR (Drive 4)

6. ✅ 8TB SMR (Drive 6)

7. ✅ 8TB SMR (Drive 7)

8. ✅ 8TB SMR (Drive 8)

 

Leave EMPTY (or minimal data):

- ⬜ 8TB SMR (Drive 9)

- ⬜ 8TB WD Red (Drive 10)

- ⬜ 4TB WD Purple (Drive 11)

- ⬜ 8TB WD Red (Drive 12)

- ⬜ 512GB SSD (Drive 13)

 

**In DrivePool, use "File Placement" rules:**

```

- Move data to fills drives 1-8 completely

- Empty drives 9, 10, 11, 12, 13

## **Phase 2: Create Initial Unraid Array**

**Shutdown Windows → Boot Unraid USB**

**Initial Unraid array (NO PARITY):**

```

Array Configuration:

├─ Cache: 512GB SSD (XFS)

├─ Disk 1: 8TB SMR (Drive 9) - WILL BE FORMATTED

├─ Disk 2: 8TB Red (Drive 10) - WILL BE FORMATTED 

├─ Disk 3: 4TB Purple (Drive 11) - WILL BE FORMATTED

└─ Disk 4: 8TB Red (Drive 12) - WILL BE FORMATTED

 

Total: 32TB usable (all formatted XFS)

Unassigned (keep NTFS, read-only):

• Drive 1-8 (all your data is here)

Format and start array - This gives you 32TB to work with.

---

Phase 3: Copy Data Stage 1

Install plugins:

bash

# In Unraid

Community Applications → Install:

- Unassigned Devices

- Dynamix File Manager (optional)

- MC (Midnight Commander) via Nerd Tools

Mount old drives:

1. Go to: Main → Unassigned Devices

2. Mount drives 1-8 as read-only

3. They appear as: /mnt/disks/WD120EDJAZ/, etc.

Copy data:

bash

# For each NTFS drive

cd /mnt/disks/[DRIVE_NAME]/PoolPart.xxx/

 

# Copy to Unraid user shares

rsync -avh --progress --ignore-existing ./ /mnt/user/media/

 

# For SMR drives (slower), use:

rsync -avh --progress --no-compress --ignore-existing ./ /mnt/user/media/

```

 

**Strategy:**

- Copy from Drive 3 (8TB SMR) first → takes ~24 hours

- Verify copy completed

- Once Drive 3 is empty, stop array

 

---

 

## **Phase 4: Expand Array (Repeat)**

 

**Add Drive 3 to array:**

1. Stop Unraid array

2. Go to Main → Array Devices

3. Assign Drive 3 (8TB SMR) to next available slot

4. Start array → formats Drive 3 as XFS

5. Now you have 40TB usable

 

**Repeat copy process:**

- Copy from Drive 4 → Add Drive 4 to array

- Copy from Drive 6 → Add Drive 6 to array

- Continue until all drives added

 

---

 

## **Phase 5: Final Unraid Configuration**

 

Once all data copied and all drives in array:

 

**Create shares with specific allocations:**

 

**1. Frigate Share (Purple drive only):**

```

Share: frigate

Included disks: disk3 (Purple 4TB)

Excluded disks: All others

```

 

**2. Documents Share (Both Reds with duplication):**

```

Share: documents

Included disks: disk2, disk4 (both 8TB Reds)

Allocation method: High-water

Minimum free space: 10GB

 

# Enable duplication at share level (not drive level)

# Use Unraid's "User Share Copy" plugin or manually rsync

```

 

**3. Media Share (Everything else):**

```

Share: media

Included disks: disk1, disk5-13 (all others)

Allocation method: Most-free

Split level: Automatically split

---

---

Order of Drive Addition (Optimized)

Copy priority (fastest first):

1. Drive 9 (8TB SMR) → already empty, format first

2. Drive 10 (8TB Red) → already empty, format first

3. Drive 11 (4TB Purple) → already empty, format first

4. Drive 12 (8TB Red) → already empty, format first

5. Then copy from and add:

6. Drive 3 (8TB SMR) → 24h copy

7. Drive 4 (8TB SMR) → 24h copy

8. Drive 6 (8TB SMR) → 24h copy

9. Drive 7 (8TB SMR) → 24h copy

10. Drive 8 (8TB SMR) → 24h copy

11. Drive 2 (10TB WD) → 30h copy

12. Drive 1 (12TB WD) → 36h copy

13. Drive 5 (18TB WD) → 48h copy (LAST - biggest)

Keep SSD (Drive 13) for cache throughout.

 

well, lots going on here. First FYI SMB on unraids doesn't have nor natively support Active Directory.

I would highly recommend going ZFS:

Since using zfs please review:
https://forums.unraid.net/topic/177887-os-70beta4-best-practice-storagediskcache-configuration/#findComment-1478045

Short answer: your plan is solid, but you can make it safer, faster, and simpler by (a) using a small ZFS or btrfs mirrored pool for your important data (docs/photos), (b) keeping Frigate on its own device, and (c) being picky about rsync flags and share settings. Below is a tight review plus a ZFS-based variant you asked for.

As I Understand it:
Quick review of your current plan

• No parity for the media: totally fine for “easy-to-replace” Plex content. Consider at least good backups for irreplaceables.

• 512 GB SSD cache: great for Docker/appdata and SABNZBD extraction. Make sure appdata, domains, and system shares are set to Primary storage: Cache (or a pool) and Mover: Prefer/Only so they don’t spray onto array disks.

• WD Purple for Frigate: ok for 24×7 sequential writes. I’d still keep Frigate’s DB/thumbs on SSD for responsiveness; keep the recordings on the Purple.

• 2×8 TB Reds “duplicated”: Unraid array itself does not do duplication like DrivePool. Use a pool (btrfs RAID1 or ZFS mirror) for this. That way duplication is automatic and bullet-proof.

• SMR drives in the array: fine for large sequential media. Don’t use them for metadata-heavy workloads.

Some of my Notes / Revisions:

Migration flow tweaks (keeps your rolling-consolidation idea)

1. Prep in Windows (DrivePool)

• Your “empty 4–5 drives, fill others” approach is good.

• Optional but recommended: run chkdsk /f on each NTFS volume before shutdown.

2. Build first Unraid boot with empty drives

• Assign the 3–4 empty disks, format XFS (or make pools—see ZFS section below). the layout you chose will matter latter I recommend pool devices only no d1 unraid array...

• Install: Community Apps → Unassigned Devices; from "unget/NerdTools" 3rd party install: screen https://slackware.pkgs.org/15.0/slackware-x86_64/screen-4.9.0-x86_64-1.txz.html.
  --Helpfull later and a nice side scripting tool... https://forums.unraid.net/topic/129200-plug-in-nerdtools/page/25/#comment-1513524

3. Copy from NTFS (Unassigned Devices) → Unraid
   Mount NTFS read-only in Unassigned Devices Plugin. Copy with rsync:

# example: NTFS drive mounted at /mnt/disks/WD10
# copy into a user share named "media"
rsync -aHAXv --info=progress2 --preallocate --numeric-ids \
      --inplace \
      /mnt/disks/WD10/PoolPart.*/ /mnt/user/media/

Notes:

• -aHAX preserves hardlinks, ACLs, xattrs (keep what’s there).

• --preallocate reduces fragmentation; --inplace avoids temp space spikes.

• For very slow SMR, add --no-compress.

• Do not mix /mnt/diskX and /mnt/user on the same copy command (classic “user share copy bug” rule). Your source is /mnt/disks so you’re safe.

4. Verify, then add the emptied drive to the array/pool
   After each copy:

# quick verification pass (no rewrites)
rsync -aHAXvn --delete /mnt/disks/WD10/PoolPart.*/ /mnt/user/media/

If the dry-run shows nothing to copy/delete, you’re good. Now stop array → assign the just-emptied disk → format → continue.

5. Permissions sweep once at the end
   Use Tools → Docker Safe New Perms (from Fix Common Problems) on your media/document shares to normalize to nobody:users.

   1. chmod 777 -R * && chown nobody:users -R *

Share layout (Unraid array + pools)

• frigate (recordings): point to the Purple disk (or a “purple” pool/dataset).

  • Include disks: only the Purple (or set Primary storage: the Purple pool).

  • Keep frigate’s DB on the SSD cache/pool (e.g., appdata/frigate).

• documents (mirrored): place this on a mirror pool (ZFS or btrfs RAID1) made from the 2×8 TB Reds.

• media (unprotected): send to the XFS array disks (all except Purple and the mirrored Reds).
  
  Review the Docs:

Manage storage | Unraid Docs

Shares | Unraid Docs

A key feature of Unraid is the management of shares. Shares are folders or drives on your Unraid server that can be accessed over a network. You can create as many shares as you want on your Unraid se

My ZFS setup Recommendation (min on Unraid 6.12+)

Unraid now supports named pools. You can create a ZFS pool in the GUI (recommended) or CLI. Here’s the practical layout:

Pools and datasets

1. Pool safe (ZFS mirror) for documents/photos

   • Devices: the two 8 TB Reds.

   • Pool properties (GUI → set per dataset after creation):

     • compression=lz4

     • xattr=sa

     • atime=off

     • recordsize=128K (good default for mixed small/medium files)

   • Datasets:

     • safe/documents (Primary storage for your “documents” share)

     • Optional: safe/photos if you want a separate share

   • Snapshots: enable regular snapshots (e.g., daily, keep 30). You can script with User Scripts:

     zfs snapshot -r safe@daily-$(date +%F)
     # prune: zfs list -t snap; zfs destroy safe@daily-YYYY-MM-DD
     

   • Off-box backups: later you can zfs send | zfs receive to another ZFS machine/NAS or to a USB backup drive with ZFS.

2. Pool purple (single-disk ZFS or XFS) for Frigate recordings

   • Device: 4 TB WD Purple.

   • If ZFS:

     • recordsize=1M (video)

     • compression=lz4

     • xattr=sa, atime=off

     • Keep sync=standard (don’t disable sync unless you fully accept data-loss risk on power loss).

   • Put Frigate clips/recordings at purple/frigate-recordings. Point the Frigate share to this dataset.

3. Pool cache (SSD, single-device)

   • Device: 512 GB NVMe/SATA SSD.

   • ZFS or btrfs are both fine here. If ZFS:

     • compression=lz4, xattr=sa, atime=off, recordsize=16K–64K are fine for DBs.

   • Datasets:

     • cache/appdata, cache/system, cache/domains

   • Set the Appdata/Systems/Domains shares: Primary storage: cache; Mover: Prefer/Only.

4. Array (unprotected XFS disks) for Plex media

   • All remaining HDDs (including SMR) as individual XFS disks, no parity.

   • Share media → Primary storage: Array; Allocation Most-free; Split level automatic.

   • Exclude the Purple and the 2×8 TB Reds from media if you want a hard separation.

Wiring shares to pools (Storage settings in each share)

• documents → Primary storage: safe (ZFS mirror). Mover: Only/Prefer safe.

• photos (if separate) → Primary storage: safe.

• frigate:

  • appdata/frigate → cache

  • frigate/recordings → purple
    In Frigate’s config, point clips/recordings to /mnt/purple/frigate-recordings (or to the share that maps there).

• media → Primary storage: Array.

Optional CLI creation (if you prefer)

Do this only once; GUI is safer and keeps Unraid in sync. (Better with zfs master plugin!)

# Mirror pool (two 8TB Reds)
zpool create -f -o ashift=12 safe mirror /dev/disk/by-id/RED_8TB_A /dev/disk/by-id/RED_8TB_B
zfs set compression=lz4 safe
zfs set xattr=sa safe
zfs set atime=off safe
zfs create safe/documents
zfs set recordsize=128K safe/documents

# Purple pool for Frigate
zpool create -f -o ashift=12 purple /dev/disk/by-id/PURPLE_4TB
zfs set compression=lz4 purple
zfs set xattr=sa purple
zfs set atime=off purple
zfs create purple/frigate-recordings
zfs set recordsize=1M purple/frigate-recordings

# Cache pool (single SSD)
zpool create -f -o ashift=12 cache /dev/disk/by-id/SSD_512G
zfs set compression=lz4 cache
zfs set xattr=sa cache
zfs set atime=off cache
zfs create cache/appdata
zfs create cache/system
zfs create cache/domains

---

Other Docker/ Frigate placement tips

• DB (SQLite) and thumbnails on SSD (/mnt/cache/appdata/frigate). (asuming one primary location for storage! per unraid shares doc...)

• Recordings on Purple (/mnt/purple/frigate-recordings), recordsize=1M.

• If you ever see dropped frames or slow clips, confirm the recordings path isn’t accidentally on the array.

?trying to setup PLex and *Arrs stack... I recommend using docker compose:
example ars compose setup...
https://forums.unraid.net/topic/194760-dockers-fail-to-communicate-on-unraid-72/#findComment-1588190

---

Preclear & health

• Before adding any drive to array/pools: run Preclear Disks (plugin) or at least a long SMART test. You’ll catch early failures.

---

Backups & “what ifs”

• Documents/Photos: ZFS mirror + snapshots + periodic send/receive to an external ZFS disk is the best “DrivePool duplication” analogue—and more robust.

• Media: you’ve accepted risk; if you later change your mind, you can add one parity and keep the rest unprotected, or move media into a ZFS pool with RAIDZ.