lyx/obsidian-yanxin
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

vault backup: 2026-03-28 21:47:21

Browse Source
This commit is contained in:
Yanxin Lu
2026-03-28 21:47:21 -07:00
parent 097265b2d3
commit 6d113980f4
1 changed files with 183 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

260
notes/Unraid Drive Replacement Plan.md
View File

@@ -1,119 +1,225 @@
# Unraid Drive Replacement Plan
## Risk Tolerance: 20% cumulative failure probability
## Goals
- Reduce array from 8 to 5 HDD drives
- Replace consumer SMR drives with enterprise CMR
- Stay under 20% cumulative failure risk before each purchase
- Minimize cost and hassle
## Current Array (as of 2026-03-28)
| Slot | Model | Serial | Age | Est. AFR | Notes |
|------|-------|--------|-----|----------|-------|
| Parity 1 | ST8000DM004 8TB SMR | ZCT2K65B | 5.3 yr | ~4.5% | Had 180 GP errors (cleared) |
| Parity 2 | ST8000DM004 8TB SMR | ZCT2K5PQ | 5.3 yr | ~3.5% | Had 19 GP errors (cleared) |
| Disk 1 | ST8000DM004 8TB SMR | ZCT2K3FF | 5.3 yr | ~2.75% | Same batch as parity drives |
| Disk 2 | ST8000DM004 8TB SMR | ZCT2K51K | 5.3 yr | ~2.75% | 1 CRC error |
| Disk 3 | ST8000DM004 8TB SMR | ZCT3XAEJ | 4.8 yr | ~3.5% | Had 10 GP errors, 2 CRC, 72K load cycles |
| Disk 4 | WD40EFRX 4TB CMR | WD-WCC7K2YF4TDJ | 6.6 yr | ~2.5% | Pristine SMART, lowest priority |
| Disk 5 | ST8000DM004 8TB SMR | ZR14W632 | 2.5 yr | ~1.75% | 56C lifetime max |
| Disk 6 | ST8000DM004 8TB SMR | ZR150CVS | 2.5 yr | ~1.75% | Clean |
| Cache | Samsung 970 EVO 1TB | — | — | — | 1% wear, excellent |
| Slot | Model | Serial | Size | Age | Est. AFR | Health |
|------|-------|--------|------|-----|----------|--------|
| Parity 1 | ST8000DM004 SMR | ZCT2K65B | 8TB | 5.3yr | ~4.5% | Had 180 GP errors (cleared) |
| Parity 2 | ST8000DM004 SMR | ZCT2K5PQ | 8TB | 5.3yr | ~3.5% | Had 19 GP errors (cleared) |
| Disk 1 | ST8000DM004 SMR | ZCT2K3FF | 8TB | 5.3yr | ~2.75% | Same manufacturing batch as parity drives |
| Disk 2 | ST8000DM004 SMR | ZCT2K51K | 8TB | 5.3yr | ~2.75% | 1 CRC error, otherwise clean |
| Disk 3 | ST8000DM004 SMR | ZCT3XAEJ | 8TB | 4.8yr | ~3.5% | Had 10 GP errors, 2 CRC, 72K load cycles |
| Disk 4 | WD40EFRX CMR | WD-WCC7K2YF4TDJ | 4TB | 6.6yr | ~2.5% | Pristine SMART, only 1.7TB used |
| Disk 5 | ST8000DM004 SMR | ZR14W632 | 8TB | 2.5yr | ~1.75% | 56C lifetime max |
| Disk 6 | ST8000DM004 SMR | ZR150CVS | 8TB | 2.5yr | ~1.75% | Clean |
| Cache | Samsung 970 EVO | S5H9NS0N831296A | 1TB | — | — | 1% wear, excellent |
All 7 Seagate drives are consumer desktop SMR (not NAS-grade). Dual parity protects against up to 2 simultaneous failures.
- Combined monthly failure rate: ~1.9%
- Hits 20% cumulative failure at ~12 months (Mar 2027)
- Total data stored: ~25TB, growing ~1TB/year
- All 7 Seagate drives are consumer desktop SMR — not NAS-grade
## Shopping List
| # | Drive | Model | Purpose | Est. Price |
|---|-------|-------|---------|------------|
| 1 | Seagate Exos X18 16TB | ST16000NM000J | Parity 1 | ~$220 |
| 2 | Seagate Exos X18 16TB | ST16000NM000J | Parity 2 | ~$220 |
| 3 | Toshiba MG08 16TB | MG08ACA16TE | Data (Disk 1) | ~$200 |
| | | | **Total** | **~$640** |
All are enterprise, CMR, 7200rpm, 5-year warranty.
## Timeline
### March 2027 — Parity 1
### Phase 0 — Now (free)
| | Details |
|--|---------|
| Remove | ST8000DM004 8TB (ZCT2K65B) |
| Install | **Seagate Exos X24 24TB** (ST24000NM002H) |
| Cost | ~$380 |
| Time | Parity sync ~18 hrs |
| Risk at swap | ~20% |
**Remove Disk 4 (WD Red 4TB)**
**Steps:** Stop array > power down > swap drive > power up > assign to Parity 1 slot > start array > parity sync
Disk 4 only has 1.7TB used. Other drives have 14.5TB of free space. No purchase needed.
Keep old drive labeled as cold spare.
1. Open **Unbalance** plugin
2. Source: Disk 4
3. Destination: Disk 3, Disk 5, Disk 6 (most free space)
4. Start transfer (~30 min for 1.7TB)
5. Verify Disk 4 is empty
6. **Stop array** (Main → Stop)
7. Unassign Disk 4 (click on the drive slot → set to "No Device")
8. **Start array** → confirm new configuration
9. Parity sync will run (~12-18 hrs, array is usable during sync)
10. Power down, physically remove the 4TB drive, power up
**Result: 8 → 7 drives.** Keep the WD Red as a cold spare — it has pristine SMART.
---
### May 2028 — Parity 2 + Disk 3
### Phase 1 — April 2027
Do these a week apart. Parity first, then data drive.
**Buy: 1x Seagate Exos X18 16TB (~$220)**
**Replace: Parity 1 (ZCT2K65B — worst drive, had 180 GP errors)**
**Remove: Disk 3 (ZCT3XAEJ — next worst, had 10 GP errors + 2 CRC + 72K load cycles)**
**Parity 2:**
Cumulative failure risk at this point: ~20%
| | Details |
|--|---------|
| Remove | ST8000DM004 8TB (ZCT2K5PQ) |
| Install | **Seagate Exos X24 24TB** (ST24000NM002H) |
| Cost | ~$380 |
| Time | Parity sync ~18 hrs |
#### Step 1: Replace Parity 1
1. **Stop array**
2. Power down server
3. Remove old Parity 1 drive (sdb, serial ZCT2K65B)
4. Install new Exos X18 16TB in the same slot
5. Power up → Main page → verify new drive shows in Parity 1 slot
6. **Start array** → Parity sync begins (~12-18 hrs)
7. Array is fully usable during sync — avoid heavy writes
8. Wait for sync to complete (check Main page for progress)
**Disk 3 (one week later):**
#### Step 2: Remove Disk 3
9. Open **Unbalance** plugin
10. Source: Disk 3 (~4.3TB)
11. Destination: Disk 1, Disk 2, Disk 5, Disk 6
12. Start transfer (~2-4 hrs)
13. Verify Disk 3 is empty
14. **Stop array**
15. Unassign Disk 3
16. **Start array** → confirm new configuration → parity sync (~12-18 hrs)
17. Power down, physically remove old Disk 3, power up
| | Details |
|--|---------|
| Remove | ST8000DM004 8TB (ZCT3XAEJ) |
| Install | **Toshiba MG10 20TB** (MG10ACA20TE) |
| Cost | ~$300 |
| Time | Rebuild ~18-20 hrs |
#### Disposal
- Old Parity 1 (ZCT2K65B): **Recycle** — worst error history
- Old Disk 3 (ZCT3XAEJ): **Recycle** — 2nd worst
Combined cost: ~$700. Risk at swap: ~20%. After this, monthly failure rate drops to ~0.9%.
**Result: 7 → 6 drives.**
---
### Mid 2030 — Disks 1 & 2
### Phase 2 — June 2027
| Slot | Install | Cost |
|------|---------|------|
| Disk 1 | **Toshiba MG10 20TB** | ~$300 |
| Disk 2 | **WD Ultrastar HC560 20TB** | ~$300 |
**Buy: 1x Seagate Exos X18 16TB (~$220)**
**Replace: Parity 2 (ZCT2K5PQ — had 19 GP errors)**
Same batch as old parity drives, will be ~9 years old. Combined: ~$600.
#### Steps
1. **Stop array**
2. Power down, swap Parity 2 with new Exos X18 16TB
3. Power up → verify drive assignment → **Start array**
4. Parity sync (~12-18 hrs)
#### Disposal
- Old Parity 2 (ZCT2K5PQ): **Recycle**
**Result: Still 6 drives.** Cannot remove another drive yet — remaining 8TB data drives are too full to consolidate without a 16TB data drive.
---
### Whenever — Disks 4, 5, 6
### Phase 3 — October 2027
Low risk, replace opportunistically on sales. ~$300 each, ~$900 total.
**Buy: 1x Toshiba MG08 16TB (~$200)**
**Replace: Disk 1 (ZCT2K3FF — same batch as old parity, 5.3yr old)**
**Remove: Disk 5 (ZR14W632 — had temperature issues)**
## Total Cost
#### Step 1: Replace Disk 1
1. **Stop array**
2. Power down, swap Disk 1 with new Toshiba MG08 16TB
3. Power up → verify drive assignment → **Start array**
4. **Rebuild** begins (~18-20 hrs) — reconstructs Disk 1's data from parity
5. Array is usable during rebuild
| Phase | When | Cost | Running Total |
|-------|------|------|--------------|
| Parity 1 | Mar 2027 | $380 | $380 |
| Parity 2 + Disk 3 | May 2028 | $700 | $1,080 |
| Disks 1 & 2 | Mid 2030 | $600 | $1,680 |
| Disks 4, 5, 6 | Whenever | $900 | $2,580 |
#### Step 2: Remove Disk 5
6. Open **Unbalance** plugin
7. Source: Disk 5 (~5.2TB)
8. Destination: Disk 1 (new 16TB, ~9.4TB free after rebuild)
9. Start transfer (~3-5 hrs)
10. Verify Disk 5 is empty
11. **Stop array**
12. Unassign Disk 5
13. **Start array** → confirm → parity sync (~12-18 hrs)
14. Power down, physically remove old Disk 5, power up
## If a Drive Fails Before Its Scheduled Replacement
#### Disposal
- Old Disk 1 (ZCT2K3FF): **Recycle** — same batch as failed parity drives
- Old Disk 5 (ZR14W632): **Keep as emergency spare** — only 2.5yr old, clean SMART
- **Parity drive fails:** Buy 1x Exos X24 24TB, swap, parity sync
- **Data drive fails:** Need 2 drives — 1x Exos X24 24TB (new parity, since 20TB data > 8TB parity) + 1x 20TB data drive. Upgrade parity first, then replace data drive.
**Result: 6 → 5 drives. Migration complete.**
Array stays online with a failed drive (dual parity). No data loss. Performance reduced until rebuild.
---
## Final Array
## Final Array (October 2027)
| Slot | Drive | Brand | Size |
|------|-------|-------|------|
| Parity 1 | Exos X24 | Seagate | 24TB |
| Parity 2 | Exos X24 | Seagate | 24TB |
| Disk 1 | MG10 | Toshiba | 20TB |
| Disk 2 | HC560 | WD | 20TB |
| Disk 3 | MG10 | Toshiba | 20TB |
| Disk 4 | TBD | TBD | 20TB+ |
| Disk 5 | HC560 | WD | 20TB |
| Disk 6 | MG10 | Toshiba | 20TB |
| Slot | Drive | Size | Used | Source | Brand |
|------|-------|------|------|--------|-------|
| Parity 1 | Exos X18 | 16TB | — | New (Apr 2027) | Seagate |
| Parity 2 | Exos X18 | 16TB | — | New (Jun 2027) | Seagate |
| Disk 1 | MG08 | 16TB | ~12TB | New (Oct 2027) | Toshiba |
| Disk 2 | ST8000DM004 | 8TB | ~6.3TB | Original | Seagate |
| Disk 6 | ST8000DM004 | 8TB | ~5.2TB | Original | Seagate |
Usable capacity: ~120TB (up from ~48TB). All CMR, all enterprise/NAS-grade, 3 brands, 5-year warranties.
- **Usable capacity: 30.6TB** (16 + 7.3 + 7.3)
- **Data stored: ~25TB** with ~6TB free
- **Growth runway: ~6 years** at 1TB/year
- **Annual failure rate: ~4%** (down from ~19%)
- **3 empty drive bays** for future expansion
## Optional Future Upgrade
When Disk 2 (5.3yr) or Disk 6 (2.5yr) eventually show wear, or if you want more space:
| Replace | With | Cost | New usable |
|---------|------|------|------------|
| Disk 2 → 16TB enterprise | Toshiba or WD | ~$200-230 | 38.6TB |
| Disk 6 → 16TB enterprise | Toshiba or WD | ~$200-230 | 48TB |
Going all-enterprise (5x 16TB) costs an additional ~$430 and gives 48TB usable (23+ years of growth). No rush — Disk 6 especially is young and healthy.
## If a Drive Fails Before Scheduled Replacement
**Parity drive fails:**
- Buy the replacement Exos X18 16TB
- Swap, parity sync, done
- Other parity still protects the array during sync
**Data drive fails (8TB):**
- Array keeps running — data is emulated from parity in real-time
- Buy a 16TB replacement (must upgrade parity first if no parity is ≥16TB yet)
- Swap, rebuild, done
**Data drive fails (new 16TB):**
- Same process, buy another 16TB
- Dual parity protects against up to 2 simultaneous failures
## Drives to Keep After Migration
| Drive | Reason |
|-------|--------|
| WD Red 4TB (WD-WCC7K2YF4TDJ) | Cold spare — pristine SMART despite 6.6yr age |
| Disk 5 8TB (ZR14W632) | Emergency spare — only 2.5yr old |
## Drives to Recycle
| Drive | Reason |
|-------|--------|
| Parity 1 (ZCT2K65B) | 180 GP error history |
| Parity 2 (ZCT2K5PQ) | 19 GP error history |
| Disk 3 (ZCT3XAEJ) | 10 GP errors + 2 CRC + 72K load cycles |
| Disk 1 (ZCT2K3FF) | Same batch as parity drives, 5.3yr |
## Pending Config Fixes
- [ ] Remove compose.manager plugin (deprecated, still installed)
- [ ] Set share security to Private (disk shares still "public")
- [ ] Move domains share to cache (useCache should be "prefer")
- [ ] Set up email/push notifications for drive alerts
- [ ] Configure UPS auto-shutdown (CyberPower CP1500, nut-dw plugin installed)
- [ ] Enable XMP in BIOS (RAM running 1333 MT/s, rated for 1600 MT/s)
These can be done anytime from the Unraid WebGUI — no downtime needed:
- [ ] Remove compose.manager plugin (Settings → Plugins → click X)
- [ ] Set disk share security to Private (Settings → Global Share Settings → Disk Shares → No)
- [ ] Set user share security to Private (Shares → each share → SMB Security → Private)
- [ ] Move domains share to cache (Shares → domains → Use Cache → Prefer, Cache Pool → cache)
- [ ] Set up email notifications (Settings → Notification Settings)
- [ ] Configure UPS auto-shutdown (Settings → UPS Settings — CyberPower CP1500 via USB, nut-dw plugin installed)
- [ ] Enable XMP in BIOS at next reboot (RAM: G.Skill DDR3 1600MHz running at 1333MHz)
- [ ] Verify appdata backups — test a restore
## SMART Test Schedule (already configured)
- Weekly short: Sundays 2:00 AM (`0 2 * * 0`)
- Monthly extended: 1st of month 3:00 AM (`0 3 1 * *`)
- Configured via User Scripts plugin