Homelab Documentation
Overview
This homelab is managed through Tailscale for secure remote access. The setup consists of a 20U rack with various servers and networking equipment optimized for virtualization, storage, and Kubernetes workloads.
Physical Infrastructure
Rack Setup
- Size: 20U rack
- Management: All hosts managed through Tailscale
- Power: Standard rack power distribution
Network Equipment
Primary Switch
- Model: Netgear JGS524e v2
- Type: 24-port Gigabit Ethernet switch
- Purpose: Main network connectivity for all rack equipment
Servers
Leviathan (Threadripper Server)
- Hardware: AMD Threadripper-based server
- Hypervisor: Proxmox VE
- Primary Purpose: Kubernetes cluster hosting via Talos VMs
- Status: Active
Virtual Machines
- seko: Arch Linux VM running on Leviathan
SuperMicro FatTwin Server
Model: SuperMicro F627R3-R72B+ 4U FatTwin Server 4-Node X9DRFR
CPU Sockets: 8x LGA2011 (2 per node)
CPU Support: Intel E5-2600v2 series
Node Configuration
Tower (Node 1)
- Hostname: tower
- OS: Unraid
- Purpose: NFS storage server
- Services:
- NFS storage for cluster
- MinIO Docker container for S3-compatible storage
- Status: Active
Melusine (Node 2)
- Hostname: melusine
- OS: Proxmox VE (installed but not running)
- Purpose: Virtualization host (planned)
- Status: Installed, not active
Node 3
- Hostname: node3
- OS: Proxmox VE (planned)
- Status: Hardware issue - CPU needs reseating
- Notes: Will install Proxmox once hardware is fixed
Node 4
- Hostname: node4
- OS: Proxmox VE (planned)
- Status: Not yet configured
Storage Configuration
Direct HDD Passthrough
For optimal storage performance on the SuperMicro nodes, direct HDD passthrough is configured via the SuperMicro RAID controller. This allows VMs to have direct access to physical drives without virtualization overhead.
Implementation: Following specialized configuration for SuperMicro hardware to bypass RAID virtualization layer.
Raspberry Pi Cluster
- Count: 3 units
- Status: Currently powered off
- Purpose: Available for lightweight services or testing
Network Architecture
Current Network Challenges
- IP address exhaustion on home network
- Need for better network segmentation
- Kubernetes cluster isolation requirements
Planned VLAN Segmentation
See Network Design & VLAN Strategy for detailed network planning.
Planned VLANs:
- VLAN 1: Management (Proxmox, switch management)
- VLAN 10: Kubernetes cluster (10.10.0.0/16)
- VLAN 20: Storage traffic (NFS, MinIO)
- VLAN 30: General services and VMs
- VLAN 40: IoT devices and RPi cluster
- VLAN 99: Guest network isolation
Management Network
All management interfaces accessible via Tailscale on VLAN 1:
- Leviathan Proxmox GUI
- Tower Unraid GUI
- Melusine Proxmox GUI (when active)
- Node 3 Proxmox GUI (planned)
- Node 4 Proxmox GUI (planned)
Service Networks
- NFS: Tower on dedicated storage VLAN
- S3 Storage: MinIO on storage VLAN
- Kubernetes: Isolated on dedicated VLAN with API access from management
Current Workloads
Active Services
- Kubernetes Cluster: Running on Talos VMs (Leviathan)
- NFS Storage: Unraid on Tower
- S3-Compatible Storage: MinIO on Tower
- Development Environment: Arch VM (seko) on Leviathan
Planned Expansions
- Additional Proxmox nodes (Melusine, Node 3, Node 4)
- Expanded virtualization capacity
- Potential Raspberry Pi cluster activation
Maintenance Notes
Pending Tasks
- Reseat CPU on Node 3
- Complete Node 4 initial setup
- Activate Melusine Proxmox installation
- Consider Raspberry Pi cluster utilization
Hardware Considerations
- SuperMicro FatTwin provides excellent density for virtualization
- Direct storage passthrough optimizes I/O performance
- Tailscale provides secure remote management without VPN complexity