homelab-docs/HARDWARE.md

Hardware Inventory

Complete hardware specifications for all homelab equipment.

Servers

PVE (10.10.10.120) - Primary Proxmox Server

CPU

• Model: AMD Ryzen Threadripper PRO 3975WX
• Cores: 32 cores / 64 threads
• Base Clock: 3.5 GHz
• Boost Clock: 4.2 GHz
• TDP: 280W
• Architecture: Zen 2 (7nm)
• Socket: sTRX4
• Features: ECC support, PCIe 4.0

RAM

• Capacity: 128 GB
• Type: DDR4 ECC Registered
• Speed: Unknown (needs investigation)
• Channels: 8-channel (quad-channel per socket)
• Idle Power: ~30-40W

Storage

OS/VM Storage:

Pool	Devices	Type	Capacity	Purpose
nvme-mirror1	2x Sabrent Rocket Q NVMe	ZFS Mirror	3.6 TB usable	High-performance VM storage
nvme-mirror2	2x Kingston SFYRD 2TB NVMe	ZFS Mirror	1.8 TB usable	Additional fast VM storage
rpool	2x Samsung 870 QVO 4TB SSD	ZFS Mirror	3.6 TB usable	Proxmox OS, containers, backups

Total Storage: ~9 TB usable

GPUs

Model	Slot	VRAM	TDP	Purpose	Passed To
NVIDIA Quadro P2000	PCIe slot 1	5 GB GDDR5	75W	Plex transcoding	Host
NVIDIA TITAN RTX	PCIe slot 2	24 GB GDDR6	280W	AI workloads	Saltbox (101), lmdev1 (111)

Total GPU Power: 75W + 280W = 355W (under load)

Network Cards

Interface	Model	Speed	Purpose	Bridge
enp1s0	Intel I210 (onboard)	1 Gb	Management	vmbr0
enp35s0f0	Intel X520 (dual-port SFP+)	10 Gb	High-speed LXC	vmbr1
enp35s0f1	Intel X520 (dual-port SFP+)	10 Gb	High-speed VM	vmbr2

10Gb Transceivers: Intel FTLX8571D3BCV (SFP+ 10GBASE-SR, 850nm, multimode)

Storage Controllers

Model	Interface	Purpose
LSI SAS2308 HBA	PCIe 3.0 x8	Passed to TrueNAS VM for EMC enclosure
Samsung NVMe controller	PCIe	Passed to TrueNAS VM for ZFS caching

Motherboard

• Model: Unknown - needs investigation
• Chipset: AMD TRX40
• Form Factor: ATX/EATX
• PCIe Slots: Multiple PCIe 4.0 slots
• Features: IOMMU support, ECC memory

Power Supply

• Model: Unknown
• Wattage: Likely 1000W+ (needs investigation)
• Type: ATX, 80+ certification unknown

Cooling

• CPU Cooler: Unknown - likely large tower or AIO
• Case Fans: Unknown quantity
• Note: CPU temps 70-80°C under load (healthy)

---

PVE2 (10.10.10.102) - Secondary Proxmox Server

CPU

• Model: AMD Ryzen Threadripper PRO 3975WX
• Specs: Same as PVE (32C/64T, 280W TDP)

RAM

• Capacity: 128 GB DDR4 ECC
• Same specs as PVE

Storage

Pool	Devices	Type	Capacity	Purpose
nvme-mirror3	2x NVMe (model unknown)	ZFS Mirror	Unknown	High-performance VM storage
local-zfs2	2x WD Red 6TB HDD	ZFS Mirror	~6 TB usable	Bulk/archival storage (spins down)

HDD Spindown: Configured for 30-min idle spindown (saves ~10-16W)

GPUs

Model	Slot	VRAM	TDP	Purpose	Passed To
NVIDIA RTX A6000	PCIe slot 1	48 GB GDDR6	300W	AI trading workloads	trading-vm (301)

Network Cards

Interface	Model	Speed	Purpose
nic1	Unknown (onboard)	1 Gb	Management

Note: MTU set to 9000 for jumbo frames

Motherboard

• Model: Unknown
• Chipset: AMD TRX40
• Similar to PVE

---

Network Equipment

UniFi Dream Machine Pro (UCG-Fiber)

• Model: UniFi Cloud Gateway Fiber
• IP: 10.10.10.1
• Ports: Multiple 1Gb + SFP+ uplink
• Features: Router, firewall, VPN, IDS/IPS
• MTU: 9216 (supports jumbo frames)
• Tailscale: Installed for VPN failover

Switches

Details needed - investigate current switch setup:

• 10Gb switch for high-speed connections?
• 1Gb switch for general devices?
• PoE capabilities?

# Check what's connected to 10Gb interfaces
ssh pve 'ip link show enp35s0f0'
ssh pve 'ip link show enp35s0f1'

---

Storage Hardware

EMC Storage Enclosure

See EMC-ENCLOSURE.md for complete details

• Model: EMC KTN-STL4 (or similar)
• Form Factor: 4U rackmount
• Drive Bays: 25x 3.5" SAS/SATA
• Controllers: Dual LCC (Link Control Cards)
• Connection: SAS via LSI SAS2308 HBA
• Passed to: TrueNAS VM (VMID 100)

Current Status:

• LCC A: Active (working)
• LCC B: Failed (replacement ordered)

Drive Inventory: Unknown - needs audit

# Get drive list from TrueNAS
ssh truenas 'smartctl --scan'
ssh truenas 'lsblk'

NVMe Drives

Model	Quantity	Capacity	Location	Pool
Sabrent Rocket Q	2	Unknown	PVE	nvme-mirror1
Kingston SFYRD	2	2 TB each	PVE	nvme-mirror2
Unknown model	2	Unknown	PVE2	nvme-mirror3
Samsung (model unknown)	1	Unknown	TrueNAS (passed)	ZFS cache

SSDs

Model	Quantity	Capacity	Location	Pool
Samsung 870 QVO	2	4 TB each	PVE	rpool

HDDs

Model	Quantity	Capacity	Location	Pool
WD Red	2	6 TB each	PVE2	local-zfs2
Unknown (in EMC)	Unknown	Unknown	TrueNAS	vault

---

UPS

Current UPS

Specification	Value
Model	CyberPower OR2200PFCRT2U
Capacity	2200VA / 1320W
Form Factor	2U rackmount
Input	NEMA 5-15P (rewired from 5-20P)
Outlets	2x 5-20R + 6x 5-15R
Output	PFC Sinewave
Runtime	~15-20 min @ 33% load
Interface	USB (connected to PVE)

See UPS.md for configuration details

---

Client Devices

Mac Mini (Hutson's Workstation)

• Model: Unknown generation
• CPU: Unknown
• RAM: Unknown
• Storage: Unknown
• Network: 1Gb Ethernet (en0) - MTU 9000
• Tailscale IP: 100.108.89.58
• Local IP: 10.10.10.125 (static)
• Purpose: Primary workstation, Happy Coder daemon host

MacBook (Mobile)

• Model: Unknown
• Network: Wi-Fi + Ethernet adapter
• Tailscale IP: Unknown
• Purpose: Mobile work, development

Windows PC

• Model: Unknown
• CPU: Unknown
• Network: 1Gb Ethernet
• IP: 10.10.10.150
• Purpose: Gaming, Windows development, Syncthing node

Phone (Android)

• Model: Unknown
• IP: 10.10.10.54 (when on Wi-Fi)
• Purpose: Syncthing mobile node, Happy Coder client

---

Rack Layout (If Applicable)

Needs documentation - Current rack configuration unknown

Suggested format:

U42: Blank panel
U41: UPS (CyberPower 2U)
U40: UPS (CyberPower 2U)
U39: Switch (10Gb)
U38-U35: EMC Storage Enclosure (4U)
U34: PVE Server
U33: PVE2 Server
...

---

Power Consumption

Measured Power Draw

Component	Idle	Typical	Peak	Notes
PVE Server	250-350W	500W	750W	CPU + GPUs + storage
PVE2 Server	200-300W	400W	600W	CPU + GPU + storage
Network Gear	~50W	~50W	~50W	Router + switches
Total	500-700W	~950W	~1400W	Exceeds UPS under peak load

UPS Capacity: 1320W Typical Load: 33-50% (safe margin) Peak Load: Can exceed UPS capacity temporarily (acceptable)

Power Optimizations Applied

See POWER-MANAGEMENT.md for details

• KSMD disabled: ~60-80W saved
• CPU governors: ~60-120W saved
• Syncthing rescans: ~60-80W saved
• HDD spindown: ~10-16W saved when idle
• Total savings: ~150-300W

---

Thermal Management

CPU Cooling

PVE & PVE2:

• CPU cooler: Unknown model
• Thermal paste: Unknown, likely needs refresh if temps >85°C
• Target temp: 70-80°C under load
• Max safe: 90°C Tctl (Threadripper PRO spec)

GPU Cooling

All GPUs are passively managed (stock coolers):

• TITAN RTX: 2-3W idle, 280W load
• RTX A6000: 11W idle, 300W load
• Quadro P2000: 25W constant (Plex active)

Case Airflow

Unknown - needs investigation:

• Case model?
• Fan configuration?
• Positive or negative pressure?

---

Cable Management

Network Cables

Connection	Type	Length	Speed
PVE → Switch (10Gb)	OM3 fiber	Unknown	10Gb
PVE2 → Router	Cat6	Unknown	1Gb
Mac Mini → Switch	Cat6	Unknown	1Gb
TrueNAS → EMC	SAS cable	Unknown	6Gb/s

Power Cables

Critical: All servers on UPS battery-backed outlets

---

Maintenance Schedule

Annual Maintenance

• Clean dust from servers (every 6-12 months)
• Check thermal paste on CPUs (every 2-3 years)
• Test UPS battery runtime (annually)
• Verify all fans operational
• Check for bulging capacitors on PSUs

Drive Health

# Check SMART status on all drives
ssh pve 'smartctl -a /dev/nvme0'
ssh pve2 'smartctl -a /dev/sda'
ssh truenas 'smartctl --scan | while read dev type; do echo "=== $dev ==="; smartctl -a $dev | grep -E "Model|Serial|Health|Reallocated|Current_Pending"; done'

Temperature Monitoring

# Check all temps (needs lm-sensors installed)
ssh pve 'sensors'
ssh pve2 'sensors'

---

Warranty & Purchase Info

Needs documentation:

• When were servers purchased?
• Where were components bought?
• Any warranties still active?
• Replacement part sources?

---

Upgrade Path

Short-term Upgrades (< 6 months)

• 20A circuit for UPS (restore original 5-20P plug)
• Document missing hardware specs
• Label all cables
• Create rack diagram

Medium-term Upgrades (6-12 months)

• Additional 10Gb NIC for PVE2?
• More NVMe storage?
• Upgrade network switches?
• Replace EMC enclosure with newer model?

Long-term Upgrades (1-2 years)

• CPU upgrade to newer Threadripper?
• RAM expansion to 256GB?
• Additional GPU for AI workloads?
• Migrate to PCIe 5.0 storage?

---

Investigation Needed

High-priority items to document:

• Get exact motherboard model (both servers)
• Get PSU model and wattage
• CPU cooler models
• Network switch models and configuration
• Complete drive inventory in EMC enclosure
• RAM speed and timings
• Case models
• Exact NVMe models for all drives

Commands to gather info:

# Motherboard
ssh pve 'dmidecode -t baseboard'

# CPU details
ssh pve 'lscpu'

# RAM details
ssh pve 'dmidecode -t memory | grep -E "Size|Speed|Manufacturer"'

# Storage devices
ssh pve 'lsblk -o NAME,SIZE,TYPE,TRAN,MODEL'

# Network cards
ssh pve 'lspci | grep -i network'

# GPU details
ssh pve 'lspci | grep -i vga'
ssh pve 'nvidia-smi -L'  # If nvidia-smi available

---

Related Documentation

• VMS.md - VM resource allocation
• STORAGE.md - Storage pools and usage
• POWER-MANAGEMENT.md - Power optimizations
• UPS.md - UPS configuration
• NETWORK.md - Network configuration
• EMC-ENCLOSURE.md - Storage enclosure details

---

Last Updated: 2025-12-22 Status: ⚠️ Incomplete - many specs need investigation