Cluster Specs 04-30-2026
+----------------------------------------------------------+
| CLUSTER TOTALS |
+----------------------+-----------------------------------+
| TOTAL CORES | 12 physical / 24 threads |
| TOTAL RAM | 192 GB DDR4 |
| TOTAL STORAGE | 1.5 TB NVMe (local only) |
| TOTAL NODES | 3 |
| TOTAL NICs | 3–7 (depending on 4-port NIC) |
| GPU | None |
| KVM/IPMI | None |
+----------------------------------------------------------+
Current Layout 04-30-2026
┌────▼────┐ ┌────▼────┐ ┌────▼────┐
│ t740 #1│ │ t740 #2│ │ t740 #3│
└────┬────┘ └────┬────┘ └────┬────┘
│ HDMI (DP→HDMI) │ HDMI (DP→HDMI) │ HDMI (DP→HDMI)
│ USB‑A │ USB‑A │ USB‑A
│ 4-port Nic PCIE │ │
└───────────────┬────────┴───────────┬─-───────────┘
│ │
▼ ▼
(All HDMI + USB cables should go a KVM I'd need to add)
(each nic must route to a Ubiquiti switch in the rack)
(Lives within 1u rack space)
Per Node Stats:
- HP t740
- AMD Ryzen V1756B
- 64gb ram DDR4 SODIMM
- 512gb NVME
- SSD M-2 sata port
- Realtek RTL8111 NIC
- PCI-E Low Profile Slot
- I have one low profile 4-port Intel GB Nic
- USB 3.1 Gen1 5Gbps Type-A
- USB 3.1 Gen2 10Gbps Type-A
- USB 3.1 Gen2 10Gbps Type-C
Possible Stack - Proxmox and Kubernetes
┌──────────────────────────────────────────────┐
│ Your Applications │
│ (AI endpoints, web apps, APIs, tools, etc.) │
└──────────────────────────────────────────────┘
┌──────────────────────────────────────────────┐
│ Kubernetes (k3s) │
│ - Deployments, services, ingress, scaling │
│ - Self-healing, rolling updates │
└──────────────────────────────────────────────┘
┌──────────────────────────────────────────────┐
│ Proxmox VMs │
│ - Each VM runs a k3s node │
│ - Snapshots, backups, migration │
└──────────────────────────────────────────────┘
┌──────────────────────────────────────────────┐
│ Proxmox Cluster │
│ - Hardware management │
│ - Storage, networking, HA │
│ - Node failover + live migration │
└──────────────────────────────────────────────┘
┌──────────────────────────────────────────────┐
│ Physical t740 Nodes │
└──────────────────────────────────────────────┘
What is missing?
- No shared storage (no live migration)
- No remote console access
- No HA for VMs
- No centralized backups
- Limited cluster networking (1GbE)
- No GPU acceleration for AI workloads
Currently Considering
- Remote Management from KVM over IP or similar from Wireguard.
- Networking between each node with USB 3 -> GB adapter or Nic.
- Physical layout within 1u spacing to optimize cooling.
- Additional storage via a NAS device or similar, if possible to place near rack.